Modeling of Airline and Passenger Dynamics in the National ... · Modeling of Airline and Passenger...

Modeling of Airline and Passenger Dynamics in the National Airspace System (NAS)

Ni Shen

Proposal of dissertation submitted to the faculty of the Virginia Polytechnic Institute and

State University in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

in

Civil and Environmental Engineering

Antonio A. Trani, Co Chair

Hojong Baik, Co Chair

Montasir M. Abbas

Raghu Pasupathy

Gerardo W. Flintsch

Aug. 27, 2010

Blacksburg, Virginia

Keywords: Agent-based model, air travel demand,

domestic leg of international passengers, multi-airport system

Copyright 2010, Ni Shen

Modeling of Airline and Passenger Dynamics in the National Airspace System

Ni Shen

Abstract

This dissertation is a collection of several models to understand airline and passenger dynamics

in the National Airspace System (NAS).

Agent-based modeling is one of the most widely used modeling simulation-analysis approaches

to understanding the dynamic behavior of complex systems. The usefulness of agent-based

modeling has been demonstrated by simulating the complex interactions between airlines,

travelers, and airports of a small-scale transportation system. Three airlines, one low cost and

two network airlines are simulated to examine how each airline behaves over time to maximize

their profit margins for a given passenger demand and operation cost structure. Passenger mode

choice and itinerary choice sub modules are embedded in the framework to characterize traveler

agent’s response to the evolved airline schedule. An airport delay model was implemented to

estimate the average delay at each airport. The estimated delay fed into the mode choice and

itinerary choice models to update the travel time related variables.

International passenger demand is a very important component of the air transportation system in

the United States. The proportion of international enplanements relative to total enplanements

increased from 8% in 1990 to 11% in 2008. Nine linear regression models are developed to

forecast the enplanements from the United States to each of nine international regions. The

international enplanements from the CONUS to each world region are modeled as a function of

GDP and GDP per capita of both the United States and the specific region. A dummy variable is

also used to account for the effects of September 11, 2001. The total number of international

enplanements is forecast to increase from 74.7 million in 2008 to 184.4 million in 2028. The

average annual growth rate is expected to be 4.7%.

The European Union – United States Open Skies Agreement, which became effective March 30,

2008. Mathematical models are developed to forecast the effect of EU-US Open Skies

Agreement on commercial airline passenger traffic over the North Atlantic Ocean. Nine

iii

econometric models were developed to forecast passenger traffic between the United States and

nine selected European countries between 2008 through 2020. 68 new nonstop flights between

the United States airports and the European airports are predicted by the model in 2020 using the

airport pair passenger demand forecast. London, Heathrow is demonstrated as an example for

rerouting the excess air travel passengers from one airport to other airports when the airport

operational capacity is exceeded.

The proportion of international enplanements relative to total enplanements within CONUS

increased from 8% in 1990 to 11% in 2008. 51% of the sampled international and U.S. territories

passengers served by U.S. carriers had at least one domestic coupon in 2007. The number of DOI

passengers through airport-pairs in each of the historical years (1990-2007) is estimated based on

the adjusted 100% international itineraries including pure international itineraries plus the non-

CONUS itineraries. The total number of DOI enplanements is estimated to grow from 37.3

million in 1990 to 79.4 million in 2007. 193 CONUS airports are estimated to have at least

10,000 DOI enplanements in 2007. The number of DOI enplanements is forecast to grow from

79.4 million in 2007 to 206.2 million in 2030 with average growth rate of 4.2% per year.

In recent years, there has been an increasing use of secondary airports both in Europe and the

U.S. Regional airports have long been considered as a possible source of relief to reduce airport

congestion at the hub airport and to efficiently accommodate future air travel demand. The

conditions under which the secondary airports develop in a metropolitan area are examined.

Fifteen multi-airport systems including 19 Operational Evolution Plan airports and 25 active

secondary airports are identified in the National Airspace System. Diverse trends of traffic

distribution among airports in the same metropolitan area are observed. We observed that the

number of markets served at the secondary airports is less than that at the primary airport in the

same metropolitan area. Most of the secondary airports are currently dominated by the low-cost

carriers. The share of seats supplied by the low-cost carriers at the secondary airports has

increased during the period 1990-2008. Full service carriers concentrate their service mainly on

the primary airport in all the multi-airport systems analyzed. The average seat capacity per

aircraft at the secondary airports is higher than that of primary airports in most of the multi-

airport systems. The secondary airports mainly serve the domestic O&D passengers.

iv

Acknowledgements

I must first thank Dr. Antonio Trani and Dr. Hojong Baik for being such good advisors to me.

Dr. Antonio Trani’s patience and humbleness along with his kindness have an incredible impact

on me. I am also grateful to Dr. Hojong Baik from whom I have received countless support and

encouragement that helped me get through many hard times. He is not only a great advisor but

also a wonderful friend. I want to extend my thanks to Dr. Raghu Pasupathy, Dr. Montasir M.

Abbas and Dr. Gerardo W. Flintsch for serving on my committee and providing valuable advice.

I am also grateful to my family who I believe is the biggest blessing in my life. It is my

grandfather, my parents and my brother’s constant supports and love that raise me up throughout

my Ph.D. work. The work is also dedicated to my grandmother who died fourteen years ago for

the remembrance of her generous love to everyone in her life.

Senanu Ashiabor, you are such a great counselor and friend to me since the first day we met. I

cannot help saying one more time that I am very thankful to God for sending an angel like you to

me. Wei Wei, Jianmin, Lu Ping, Jianxia and Fu Zhuo, your wonderful friendships make me

believe God’s mercy and love to me is beyond what I can imagine. I have to thank Jessie for her

considerate advice that helps me make several important decisions in Blacksburg.

Finally, I would like to thank Howard Swingle in my lab and my roommate Gillian Boulay for

their helpful advice and comments on how to improve professional writing.

v

Table of Contents

1 Introduction ............................................................................................................................. 1 1.1 An Agent-Based Model of Airline Evolution .................................................................. 1 1.2 International Enplanements within the Continental U.S. ................................................. 2 1.3 The Impact of the EU-US Open Skies Agreement on Commercial Airline Passenger Traffic over the North Atlantic ................................................................................................... 2 1.4 Domestic Leg of International Passengers within the Continental U.S. .......................... 3 1.5 Development of Multi-Airport System and Commercial Airline Network in the National Airspace System (NAS) .............................................................................................................. 3

2 An Agent-Based Model of Airline Evolution (to be submitted to Journal of Air Transport Management) .................................................................................................................................. 4

2.1 Introduction ...................................................................................................................... 5 2.1.1 Agent-based model ................................................................................................... 5 2.1.2 Motivation ................................................................................................................. 6 2.1.3 Objective ................................................................................................................... 7

2.2 Literature Review ............................................................................................................. 7 2.3 Methodology .................................................................................................................. 10

2.3.1 Overview of the ABM Model ................................................................................. 10 2.3.2 Traveler Agent Behavior......................................................................................... 13 2.3.3 Airline Agent Behavior ........................................................................................... 18 2.3.4 Airport Agent Behavior .......................................................................................... 25

2.4 Simulation Results.......................................................................................................... 27 2.5 Conclusions and Recommendations ............................................................................... 32 2.6 References ...................................................................................................................... 33

3 International Enplanements within the Continental U.S. (to be submitted to Journal of Air Transport Management) ............................................................................................................... 36

3.1 Introduction .................................................................................................................... 37 3.2 Literature Review ........................................................................................................... 40

3.2.1 Modeling of Air Travel Demand ............................................................................ 40 3.2.2 Modeling of Air Travel Demand by FAA, Airbus and Boeing .............................. 49

3.3 Methodology .................................................................................................................. 51 3.3.1 Scatterplot and Correlation Matrix ......................................................................... 54 3.3.2 Assumed Functional Form of the Model ................................................................ 55 3.3.3 Model Evaluation .................................................................................................... 56

vi

3.4 Results ............................................................................................................................ 62 3.4.1 Model Development and Evaluation Results .......................................................... 62 3.4.2 Forecast Results of Passenger Enplanements from the U.S. to each International Region ................................................................................................................................. 68 3.4.3 Evaluation of the Forecast Results .......................................................................... 71 3.4.4 Model Application .................................................................................................. 71

3.5 Conclusion ...................................................................................................................... 72 4 The Impact of the EU-US Open Skies Agreement on Commercial Airline Passenger Traffic over the North Atlantic (to be submitted to Journal of Air Transport Management) .................. 74

4.1 Introduction .................................................................................................................... 76 4.2 Domain of Analysis ........................................................................................................ 76

4.2.1 Time Frame of Forecast Model ............................................................................... 76 4.2.2 Geographical Domain of Analysis .......................................................................... 77 4.2.3 Airports Included in Domain of Analysis ............................................................... 81

4.3 Modeling Demand among the United States and Selected European Countries ........... 86 4.3.1 Models Recommended............................................................................................ 86 4.3.2 Other Models Investigated ...................................................................................... 90

4.4 Distribution of United States to European Country Passenger Demand to Airport Pairs .. ........................................................................................................................................ 95 4.5 New Nonstop Flights Suggested by Forecast ............................................................... 101 4.6 Criteria for suggesting new nonstop flights ................................................................. 103 4.7 Demand for travel between gateways without nonstop flights .................................... 107 4.8 Demonstration of Capacity Constraints ....................................................................... 113 4.9 Demonstration of Method with London, Heathrow ..................................................... 116 4.10 Conclusions and Recommendations ......................................................................... 118

5 Domestic Leg of International Passengers within the Continental U.S. (to be submitted to Journal of Air Transport Management) ...................................................................................... 120

5.1 Introduction .................................................................................................................. 122 5.2 Data sources ................................................................................................................. 126

5.2.1 T-100 data ............................................................................................................. 126 5.2.2 Airline O&D Survey (DB1A/DB1B) .................................................................... 128

5.3 Methodology ................................................................................................................ 130 5.3.1 Estimation of DOI Passengers through Airport-Pairs during 1990-2007 ............. 131

5.4 Forecast of DOI during 2008-2030 .............................................................................. 137 5.5 Results .......................................................................................................................... 140

vii

5.5.1 The Total Number of DOI Enplanements ............................................................. 140 5.5.2 The Number of DOI Enplanements at Airports .................................................... 140 5.5.3 The Number of DOI Passengers through Airport-Pairs ........................................ 142

5.6 Conclusions .................................................................................................................. 144 6 Development of Secondary Airports in Multi-Airport Systems The Impact of the EU-US Open Skies Agreement on Commercial Airline Passenger Traffic over the North Atlantic in the NAS (to be submitted to Journal of Air Transport Management) .............................................. 145

6.1 Introduction .................................................................................................................. 147 6.2 Literature Review ......................................................................................................... 149

6.2.1 Dynamics of Evolution of Multi-Airport System ................................................. 149 6.2.2 Airport Choice Models in Multi-Airport System.................................................. 150

6.3 Identification of Multi-Airport System in the NAS ..................................................... 155 6.4 Markets served at secondary airports in MAS ............................................................. 160 6.5 Traffic Distribution by Markets and Carrier at the Secondary Airports in MAS ........ 164 6.6 Aircraft Type at the Secondary Airports in MAS ........................................................ 169 6.7 Flight Distance at the Secondary Airports in MAS ...................................................... 172 6.8 Connectivity at the Secondary Airports in MAS ......................................................... 175 6.9 Conclusion .................................................................................................................... 179 6.10 Reference .................................................................................................................. 181

7 Appendix ............................................................................................................................. 184 Appendix A: An Agent-Based Model of Airline Evolution .................................................. 185

A.1 Evolution of Airline Profit, Load Factor, Airfare, Aircraft Size, Market Share for the 1st Scenario (Adjust Airfare and Aircraft size with Fuel Price: $1.5/gal) ............................... 186 A.2 Evolution of Airline Profit, Load Factor, Airfare, Aircraft Size, Market Share for the 2nd Scenario (Adjust Airfare and Aircraft size with Fuel Price: $2.0/gal) ........................ 188 A.3 Evolution of Airline Profit, Load Factor, Airfare, Aircraft Size, Market Share for the 3rd Scenario (Adjust Airfare, Aircraft size and Add/Cancel Flights with Fuel Price: $1.5/gal) .............................................................................................................................. 190 A.4 Input Data for the Model .............................................................................................. 192 A.5 Other Predefined Input Variables ................................................................................ 205

Appendix B: International Enplanements within the Continental U.S. (CONUS) ................. 206 B.1 Countries Covered by each World Region ................................................................... 207 B.2 Historical (1990 – 2008) and Forecast (2009 – 2040) Enplanements within the CONUS to Each of the Nine International Regions .......................................................................... 211 B.3 Comparison between Historical and Forecast Enplanements within the CONUS to Each of the Nine International Regions during 1990 – 2008 ....................................................... 217

viii

B.4 Input Data for the Model .............................................................................................. 222 B.5 Summary of Matlab Functions and Data Preparation .................................................. 232 B.6 Code of Matlab Functions ............................................................................................ 240

Appendix C: The Impact of the EU-US Open Skies Agreement on Commercial Airline Passenger Traffic over the North Atlantic .............................................................................. 251

C.1 Passenger Traffic Symmetry Observed at the Top 91 Airport Pairs between the United States and the Selected Nine European Countries .............................................................. 252 C.2 Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the United States to Each of the Selected Nine European Countries ................................................... 253 C.3 Comparison between Historical and Forecast Enplanements from the U.S. to Each of the Selected Nine European Counties during 1990 – 2007 ................................................ 261 C.4 Input Data for the Model .............................................................................................. 266

Appendix D: Domestic Leg of International Passengers within the Continental U.S. (CONUS)................................................................................................................................................. 273

D.1 Summary of Matlab Functions ..................................................................................... 274 D.2 Description of Input and Output Variables for Matlab Functions ............................... 281

Appendix E: Development of Secondary Airports in Multi-Airport Systems (MAS) in the NAS......................................................................................................................................... 283

E.1 Evolution of Enplanements Share at Airports in Multi-Airport System ...................... 284 E.2 Non-Stop Markets Served at Airports in Multi-Airport System & Traffic Distribution by Market/Carrier among Airports in Multi-Airport System .................................................. 289 E.3 Traffic Distribution by Market & Carrier among Airports in Boston Multi-Airport System ................................................................................................................................. 296 E.4 Average Flight Distance at Airports in Multi-Airport System ..................................... 299 E.5 Connectivity at Secondary Airports in Multi-Airport System ..................................... 301 E.6 Summary of Matlab Functions ..................................................................................... 302 E.7 Description of Input and Output Variables for Matlab Functions ................................ 307

ix

List of Figures

Figure 2-1: Interactions between Airlines, Airports and Travelers in the National Transportation

System. ............................................................................................................................................ 6

Figure 2-3: Flow Chart of Airline Strategy Evolution Process. .................................................... 12

Figure 2-4: Flowchart for Itinerary Choice Model. ...................................................................... 16

Figure 2-5: Hypothetic Airline Network and Distance (Miles) Between Airports. ...................... 18

Figure 2-6: Main Variables and Their Intersections. .................................................................... 26

Figure 2-7: Comparison of Airline Profit & Load Factor between 1st and 2nd Scenario .............. 28

Figure 2-8: Comparison of Market Share of Airlines & Aggregate Market Share of Commercial

Air between 1st and 2nd Scenario. .................................................................................................. 28

Figure 2-9: Comparison of Evolutions of Airline Average Airfare & Seat capacity per Flight

between 1st and 2nd Scenario. ........................................................................................................ 29

Figure 2-10: Airline Profit, Load Factor, Airfare and Seat capacity of 3rd Scenario with Evolving

Airfare, Aircraft Size and Canceling / Adding Flights. ................................................................ 30

Figure 2-11: Comparison of Airline Equilibrium Profit & Load Factor between 1st (right plots)

and 3rd (left plots) Scenario. .......................................................................................................... 30

Figure 2-12: Evolution of Airline Schedule and Performance Statistics (Left plots: 1st Scenario*;

Middle plots: 2nd Scenario*; Right plots: 3rd Scenario*). * 1st Scenario: Adjust Airfare and

Aircraft size with Fuel Price: $1.5/gal; *2nd Scenario: Adjust Airfare and Aircraft size with Fuel

Price: $2.0/gal; *3rd Scenario: Adjust Airfare, Aircraft size and Add/Cancel Flights with Fuel

Price: $1.5/gal; .............................................................................................................................. 31

Figure 3-1: Evolution of International Enplanements/Seats/Departures within CONUS (Data

Source: 1990 - 2007 T100 International Market Data)................................................................ 38

Figure 3-2: Evolution of Share of International Enplanements/Seats/Departures over

Corresponding Total Values within CONUS (Data Source: 1990 - 2007 T100 International

Market Data). ................................................................................................................................ 39

Figure 3-3: Share of International Enplanements over Total Enplanements at 15 Airports with

Most International Enplanements (Data Source: 1990 - 2007 T100 Domestic/International

Market Data). ................................................................................................................................ 39

x

Figure 3-4: Enplanements within the CONUS to each of 12 Regions in the analysis (Data

Source: 2008 T100 International Market Data). .......................................................................... 52

Figure 3-5: Evolution of Enplanements within the CONUS to each of the Nine International

Regions (Data Source: 1990-2008 T100 International Market Data). ........................................ 53

Figure 3-6: Methodology Framework Used to Forecast International Air Travel Demand. ........ 54

Figure 3-7: Normal Probability Plot of Regression Standardized Residual (Normality Test -

Model to Forecast Air Passengers from the U.S. to Europe). ....................................................... 66

Figure 3-8: Histogram of Regression Standardized Residual (Zero Mean Test - Model to

Forecast Air Passenger from the U.S. to Europe). ........................................................................ 66

Figure 3-9: Regression Standardized Residual V.S. Standardized Predicted Value

(Homoscedasticity Test - Model to Forecast Air Passenger from the U.S. to Europe). ............... 67

Figure 3-10: Comparison between Historical and Forecast Enplanements within the CONUS to

Europe during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data). 68

Figure 3-11: Comparison Between Forecast Growth and Historical Growth. .............................. 69

Figure 3-12: Historical (1990 – 2008) and Forecast (2008 – 2040) Enplanements within the

CONUS to Europe (Historical Source: 1990 - 2008 T100 International Market Data). ............. 70

Figure 3-13: Historical (1990 – 2008) and Forecast (2009 – 2040) Enplanements within the

CONUS to All 12 Regions (Historical Source: 1990 - 2008 T100 International Market Data). 70

Figure 3-14: Evolution Framework of Transportation System Analysis Model (TSAM)............ 72

Figure 4-1: 2007 Passenger Traffic from the United States to European Countries (Source: 2007

T100 International Market Data). ................................................................................................ 77

Figure 4-2: 2007 Cumulative Percentage of Total Passenger Traffic from the U.S. to Europe

(Source: 2007 T100 International Market Data).......................................................................... 78

Figure 4-3: Map of Europe and Nine Selected European Countries in the analysis. .................... 79

Figure 4-4: Country-to-Country Passenger Traffic from Europe to the United States vs. Traffic

from the United States to Europe (Source: 2007 T100 International Market Data). ................... 80

Figure 4-5: Airport-to-Airport Passenger Traffic from Europe to the United States vs. Traffic

from the United States to Europe (Source: 2007 T100 International Market Data). ................... 80

Figure 4-6: 1990 – 2007 Historical Passenger Traffic from the United States to Denmark

(Source: 2007 T100 International Market Data).......................................................................... 81

xi

Figure 4-7: 2007 Passenger Traffic from Selected 31 United States Airports in the analysis

Domain to All the Selected 35 European Airports (Source: 2007 T100 International Market

Data). ............................................................................................................................................ 82

Figure 4-8: 2007 Passenger Traffic from All the Selected 31 United States Airports to each 35

Selected European Airports in the analysis Domain (Source: 2007 T100 International Market

Data). ............................................................................................................................................ 83

Figure 4-9: 2007 Selected 31 United States Airports’ Cumulative Percentage of Total Passenger

Traffic from the United States to Europe (Source: 2007 T100 International Market Data). ...... 84

Figure 4-10: 2007 Selected 35 European Airports’ Cumulative Percentage of Total Passenger

Traffic from the United States to Europe (Source: 2007 T100 International Market Data). ...... 85

Figure 4-11: Comparison between Historical and Forecast Enplanements from the U.S. to U.K.

during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market Data). ............ 88

Figure 4-12: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to

United Kingdom (Historical Source: 1990 - 2007 T100 International Market Data). ................ 89

Figure 4-13: 1998 – 2007 Scatter Plot of Nature Logarithm of Passenger Traffic from the United

States to Selected Nine European Countries and Nature Logarithm of Product of United States’

GDP and European Country’s GDP (Passenger Traffic Source: 1998 - 2007 T100 International

Market Data; GDP Source: 1998 – 2007 USDA International Macroeconomic Data Set). ...... 92

Figure 4-14: 1998 – 2007 Scatter Plot of Nature Logarithm of Passenger Traffic from the United

States to Selected Nine European Countries and Nature Logarithm of Associated Average

Airfare from United States to Nine European Countries (Passenger Traffic Source: 1998 - 2007

T100 International Market Data; Airfare Source: 1998 – 2007 DB1B Data). ............................ 93

Figure 4-15: 1998 – 2007 Average Airfare from the United States to Selected Nine European

Countries (Data Source: 1998 - 2007 DB1B Data). ..................................................................... 94

Figure 4-16: 1998 – 2007 Passenger Traffic from the United States to Selected Nine European

Countries (Data Source: 1998 - 2007 T100 International Market Data). .................................... 95

Figure 4-17: Spreadsheet illustration of Fratar Method and Iterative Process for Distributing

United States to Europe Passengers to Airport Pairs in Future Years. ....................................... 100

Figure 4-18: Distribution of the number of airlines providing nonstop service (Source: 2007

T100I Segment Data). ................................................................................................................. 104

xii

Figure 4-19: Distribution of the number of months for single airline to provide nonstop service

(Source: 2007 T100I Segment Data). ......................................................................................... 105

Figure 4-20: Passengers per airport pair per year for the 89 airport pairs provided with nonstop

flight by a single airline (Source: 2007 T100I Segment Data). .................................................. 106

Figure 4-21: Flight frequency per airport pair per year for the 89 airport pairs 89 airport pairs

provided with nonstop flight by a single airline (Source: 2007 T100I Segment Data). ............. 107

Figure 4-22: 104 Unique Routes for Passengers Travelling through ATL and LHR using ORD

and LHR airport pair where nonstop service is provided (Source: 2007 DB1B Data). ............. 108

Figure 5-1: Domain of Domestic Enplanements Due to International or Non-CONUS Itineraries

(DOI Enplanements). .................................................................................................................. 122

Figure 5-2: Share of Sampled DOI Enplanements over Total Sampled Domestic Enplanements at

15 Airports with Most DOI Enplanements in 2007 (Data Source: 2007 DB1B Data). ............. 125

Figure 5-4: _ 1 for Thirty-Nine Gateway Airports for Pure International Itineraries.

..................................................................................................................................................... 135

Figure 5-5: _ 2 for Thirty-Nine Gateway Airports for Pure International Itineraries

in 2007. ....................................................................................................................................... 136

Figure 5-6: _ 1 for Thirty-Three Gateway Airports for the U.S. Territory Travel. 137

Figure 5-7: Thirty-Three Gateway Airports for the U.S. Territory Travel. ................................ 138

Figure 5-8: Top 30 Airports’ Share of Total DOI Enplanements within the CONUS in 2007. . 139

Figure 5-9: Historical (1990-2007) and Forecast (2008-2030) DOI Enplanements within the

CONUS. ...................................................................................................................................... 140

Figure 5-10: Twenty-Three Airports with More Than One Million DOI Enplanements within the

CONUS in 2007. ......................................................................................................................... 141

Figure 5-11: Historical (1990-2007) and Forecast (2008-2030) DOI Enplanements at the Top 5

Airports. ...................................................................................................................................... 142

Figure 5-12: Thirty CONUS Flight Coupons with Most International & Non-CONUS Passenger

Traffic in 2007. ........................................................................................................................... 143

Figure 5-13: Historical (1990-2007) and Forecast (2008-2030) of Sum of International or Non-

CONUS Passenger Traffic in Two Directions for Top 5 Airport-Pairs. ..................................... 144

The previous models of airport choice in multi-airport system are summarized in Table 6-1. .. 151

xiii

Figure 6-1: Number of Secondary Airports for 34 OEP Airports in the CONUS (Source: 1990-

2008 T-100 Segment). ................................................................................................................. 155

Figure 6-2: Evolution of Enplanements at Secondary (Active vs. Inactive) Airports in Boston

Area (Source: 1990-2008 T-100 Segment). ................................................................................ 156

Figure 6-3: Evolution of Enplanements Share at Secondary Airports (Source: 1990-2008 T-100

Segment). ..................................................................................................................................... 158

Figure 6-4: Evolution of Enplanements Share at Active Airports in Boston Area (Source: 1990-

2008 T-100 Segment). ................................................................................................................. 159

Figure 6-5: Evolution of Enplanements Share at Active Airports in Houston Area (Source:

1990-2008 T-100 Segment). ........................................................................................................ 159

Figure 6-6: Evolution of Enplanements Share at Active Airports in Miami Area (Source: 1990-

2008 T-100 Segment). ................................................................................................................. 160

Figure 6-7: Markets Served at Airports in Multi-Airport System of San Francesco (Source: 2008

OAG). .......................................................................................................................................... 161

Figure 6-8: Seat capacity by Market Served at Airports in San Francesco Bay Area (Source:

2008 OAG). ................................................................................................................................. 162

Figure 6-9: Seat capacity by Carrier at Airports in San Francesco Bay Area (Source: 2008

OAG). .......................................................................................................................................... 164

Figure 6-10: LCCs Share of Enplanements at 44 Airports in 15 Multi-Airport Systems in the

NAS (Source: 1990, 2000, 2008 T-100 Segment). ..................................................................... 165

Figure 6-11: “Southwest Effect” at PVD (Boston/Providence) and MHT (Boston/Manchester)

(Source: 1990-2008 T-100 Segment). ......................................................................................... 166

Figure 6-12: Markets Served by Southwest at Airports in MAS of Los Angeles & Boston

(Source: 2008 OAG). .................................................................................................................. 167

Figure 6-13: Markets Served by Full Service Carriers at Airports in Boston Area (Source:

2008 OAG). ................................................................................................................................. 169

Figure 6-14: Distribution of Aircraft Type at 44 Airports in 15 Multi-Airport Systems in the

NAS (Source: 2008 OAG). ......................................................................................................... 171

Figure 6-15: Distribution of Flight Range at 44 Airports in 15 Multi-Airport Systems in the NAS

(Source: 2008 OAG). .................................................................................................................. 173

xiv

Figure 6-16: Connecting Rate at 44 Airports in 15 Multi-Airport Systems in the NAS (Source:

2008 Domestic DB1B). ............................................................................................................... 176

Figure 6-17: Connecting Markets Covering 90% of Connecting Passengers at MDW and HOU

(Source: 2008 Domestic DB1B). ................................................................................................. 177

Figure 6-18: OD Airport-Pairs Covering 90% of Connecting Passengers at MDW and HOU

(Source: 2008 Domestic DB1B). ................................................................................................. 178

Figure B-1: Historical (1990 – 2008) and Forecast (2009 – 2040) Enplanements within the

CONUS to Africa (Historical Source: 1990 - 2008 T100 International Market Data). ............ 211

Figure B-2: Historical (1990 – 2008) & Forecast (2009 – 2040) Enplanements within the

CONUS to Asia (Historical Source: 1990 - 2008 T100 International Market Data). ............... 211


CONUS to Canada (Historical Source: 1990 - 2008 T100 International Market Data). .......... 212


CONUS to Caribbean & Central America (Historical Source: 1990 - 2008 T100 International

Market Data). .............................................................................................................................. 212


CONUS to Europe (Historical Source: 1990 - 2008 T100 International Market Data). ........... 213


CONUS to Mexico (Historical Source: 1990 - 2008 T100 International Market Data). .......... 213


CONUS to Middle East (Historical Source: 1990 - 2008 T100 International Market Data). ... 214


CONUS to Oceania (Historical Source: 1990 - 2008 T100 International Market Data). ......... 214


CONUS to South America (Historical Source: 1990 - 2008 T100 International Market Data).

..................................................................................................................................................... 215

Figure B-10: Comparison between Historical and Forecast Enplanements within the CONUS to

Africa during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).

..................................................................................................................................................... 217


Asia during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data). . 217

xv


Canada during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).

..................................................................................................................................................... 218


Caribbean & Central America during 1990 – 2008 (Historical Source: 1990 - 2008 T100

International Market Data)......................................................................................................... 218


Europe during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).

..................................................................................................................................................... 219


Mexico during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).

..................................................................................................................................................... 219


Middle East during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market

Data). .......................................................................................................................................... 220


Oceania during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).

..................................................................................................................................................... 220


South America during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market

Data). .......................................................................................................................................... 221

Figure C-1: Airport-to-Airport Passenger Traffic from Europe to the United States vs. Traffic

from the United States to Europe (Source: 2007 T100 International Market Data). ................. 252

Figure C-2: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to

the U.K. (Historical Source: 1990 - 2007 T100 International Market Data). .......................... 253


Germany (Historical Source: 1990 - 2007 T100 International Market Data). ......................... 253


France (Historical Source: 1990 - 2007 T100 International Market Data). ............................. 254


Netherlands (Historical Source: 1990 - 2007 T100 International Market Data). ..................... 254

xvi


Italy (Historical Source: 1990 - 2007 T100 International Market Data). ................................. 255


Ireland (Historical Source: 1990 - 2007 T100 International Market Data). ............................. 255


Spain (Historical Source: 1990 - 2007 T100 International Market Data). ............................... 256


Switzerland (Historical Source: 1990 - 2007 T100 International Market Data). ..................... 256

Figure C-10: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S.

to Belgium (Historical Source: 1990 - 2007 T100 International Market Data). ...................... 257

Figure C-11: Comparison between Historical and Forecast Enplanements from the U.S. to the

U.K. during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market Data. . 261

Figure C-12: Comparison between Historical and Forecast Enplanements from the U.S. to

Germany during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market Data).

..................................................................................................................................................... 261

Figure C-13: Comparison between Historical and Forecast Enplanements from the U.S. to France

during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market Data). .......... 262


Netherlands during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market

Data). .......................................................................................................................................... 262

Figure C-15: Comparison between Historical and Forecast Enplanements from the U.S. to Italy



Ireland during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market Data).

..................................................................................................................................................... 263

Figure C-17: Comparison between Historical and Forecast Enplanements from the U.S. to Spain



Switzerland during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market

Data). .......................................................................................................................................... 264

xvii


Belgium during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market Data).

..................................................................................................................................................... 265

Figure C-20: Enplanements from the United States to the Selected Nine European Countries in

Analysis Domain during 1990 - 2007 (Source: 1990 - 2007 T100 International Market Data).266

Figure E-1: Evolution of Enplanements Share at Airports in the New York City MAS (Source:

1990-2008 T-100 Segment). ........................................................................................................ 284

Figure E-2: Evolution of Enplanements Share at Airports in the Chicago MAS (Source: 1990-

2008 T-100 Segment). ................................................................................................................. 284

Figure E-3: Evolution of Enplanements Share at Airports in the Los Angeles MAS (Source:

1990-2008 T-100 Segment). ........................................................................................................ 285

Figure E-4: Evolution of Enplanements Share at Airports in the Dallas MAS (Source: 1990-

2008 T-100 Segment). ................................................................................................................. 285

Figure E-5: Evolution of Enplanements Share at Airports in the Washington D.C. MAS (Source:

1990-2008 T-100 Segment). ........................................................................................................ 285

Figure E-6: Evolution of Enplanements Share at Airports in the San Francesco MAS (Source:

1990-2008 T-100 Segment). ........................................................................................................ 286

Figure E-7: Evolution of Enplanements Share at Airports in the Orlando MAS (Source: 1990-

2008 T-100 Segment). ................................................................................................................. 286

Figure E-8: Evolution of Enplanements Share at Airports in the Detroit MAS (Source: 1990-

2008 T-100 Segment). ................................................................................................................. 286

Figure E-9: Evolution of Enplanements Share at Airports in the Philadelphia MAS (Source:

1990-2008 T-100 Segment). ........................................................................................................ 287

Figure E-10: Evolution of Enplanements Share at Airports in the Tampa MAS (Source: 1990-

2008 T-100 Segment). ................................................................................................................. 287

Figure E-11: Evolution of Enplanements Share at Airports in the Cincinnati MAS (Source:

1990-2008 T-100 Segment). ........................................................................................................ 287

Figure E-12: Evolution of Enplanements Share at Airports in the Cleveland MAS (Source:

1990-2008 T-100 Segment). ........................................................................................................ 288

Figure E-13: Evolution of Enplanements at Airports in the Houston MAS (Source: 1990-2008

T-100 Segment). .......................................................................................................................... 288

xviii

Figure E-14: Non-Stop Markets Served in the New York City MAS – JFK* (Kennedy), EWR*

(Newark), LGA* (LarGuardia), ISP (Islip), HPN (Westchester), SWF (Stewart) (The airport

marked by * is the primary airport. This applys to the rest of this section). .............................. 289

Figure E-15: Non-Stop Markets Served in the Chicago MAS. .................................................. 290

Figure E-16: Traffic Distribution by Market/Carrier among Airports in the Chicago MAS. ..... 290

Figure E-17: Non-Stop Markets Served in the Los Angeles MAS – LAX* (Int’l), SNA (Orange

County), ONT (Ontario), BUR (Burbank), LGB (Long Beach). ................................................ 291

Figure E-18: Non-Stop Markets Served in the Dallas MAS ....................................................... 292

Figure E-19: Traffic Distribution by Market/Carrier among Airports in the Dallas MAS. ........ 292

Figure E-20: Non-Stop Markets Served in the Houston MAS ................................................... 293

Figure E-21: Traffic Distribution by Market/Carrier among Airports in Houston MAS. .......... 293

Figure E-22: Non-Stop Markets Served in the Miami MAS – MIA* (Int’l), FLL* (Fort

Lauderdale), PBI (Palm Beach). ................................................................................................. 294

Figure E-23: Non-Stop Markets Served in the Boston MAS ..................................................... 295

Figure E-24: Traffic Distribution by Market/Carrier among Airports in the Boston MAS. ....... 295

Figure E-25: Traffic Distribution by Market for FSCs 1 among Airports in MAS of Boston. .. 296

Figure E-26: Traffic Distribution by Market for FSCs 2 among Airports in MAS of Boston. .. 297

Figure E-27: Traffic Distribution by Market for LCCs among Airports in MAS of Boston. ..... 298

Figure E-28: Average Flight Distance at 44 Airports in 15 Multi-Airport Systems in the NAS

(Source: 2008 OAG). .................................................................................................................. 299

Figure E-29: Southwest Airline Average Flight Distance during 1990-2008 (Source: 1990-2008

T-100 Segment). .......................................................................................................................... 300

Figure E-30: Southwest Airline Short-, Medium-haul Flight Share during 1990-2008 (Source:

1990-2008 T-100 Segment). ........................................................................................................ 300

Figure E-31: Top 5 OD Airport-Pairs of Connecting Passengers at MDW (Chicago/Midway) 301

Figure E-32: Top 5 OD Airport-Pairs of Connecting Passengers at DAL (Dallas/Love-Field) . 301

Figure E-33: Top 5 OD Airport-Pairs of Connecting Passengers at HOU (Houston/Hobby). ... 301

xix

List of Tables

Table 2-1: Summary of agent-based models simulating the key participants in the National

Transportation System (NTS). ........................................................................................................ 8

Table 2-2: Airline Characteristics. ................................................................................................ 18

Table 2-3: Aircraft Characteristics. ............................................................................................... 19

Table 2-4: The Predefined Order and Frequency Airline Agents Use to Evolve the Strategies. . 22

Table 2-5: An Example to Show the Rules Used by Airline Agents to Evolve Strategies. .......... 23

Table 3-1: Summary of Previous Passenger Demand Models. ..................................................... 42

Table 3-2: Socioeconomic Factors, Chosen Function Forms and Assigned Notations of Demand

Variables (Rengaraju and Thamizh Arasan, 1992). ...................................................................... 47

Table 3-3: Definition of 12 Regions by World Area Code (WAC). ............................................. 51

Table 3-4: Regression Equations to Forecast Air Passenger Demand from the U.S. to each of the

Nine International Regions. .......................................................................................................... 64

Table 3-5: Adjustment Factors for Estimated International Air Travel Demand Models. ........... 68

Table 4-1: Statistical Metrics for Semi-Log Country-to-Country Passenger Demand Model. .... 87

Table 4-2: Adjustment Factors for Semi-Logarithmic Country-to-Country Passenger Demand

Model. ........................................................................................................................................... 89

Table 4-3: Estimated Coefficients for the Fixed Effect Model Using Weighted Least Squares

Regression. .................................................................................................................................... 93

Table 4-4: New Nonstop Transatlantic Airport Pairs and Their Forecast Passenger Demand in

Year 2010, 2015 and 2020. ......................................................................................................... 102

Table 4-4: New Nonstop Transatlantic Airport Pairs and Their Forecast Passenger Demand in

Year 2010, 2015 and 2020 (Continued). ..................................................................................... 103

Table 4-5: Process for Estimating Total Passenger Travelling between Atlanta and London,

Heathrow. .................................................................................................................................... 110

Table 4-6: 2010 Passenger Traffic for 871 Transatlantic Airport Pairs without Nonstop Service.

..................................................................................................................................................... 111

Table 4-6: 2010 Passenger Traffic for 871 Transatlantic Airport Pairs without Nonstop Service

(Continued). ................................................................................................................................ 112

xx

Table 4-7: Estimated DB1B Connecting Passenger Ratio at 35 European Airports in Domain in

Base Year (2007) (Source: 2007 DB1B Data). .......................................................................... 115

Table 4-8: Estimated Connecting Passenger at Top 13 European Connecting Airports in Base

Year (2007) (Source: 2007 DB1B Data; 2007 T100 Market Data). .......................................... 116

Table 4-9: Assignment of the excess terminating passengers at London, Heathrow to the

candidate terminating airports..................................................................................................... 117

Table 4-10: Assignment of the excess connecting passengers at London, Heathrow to the

candidate connecting airports. .................................................................................................... 118

Table 4-11: Overview of Assignment of the excess passengers at London, Heathrow to the

candidate connecting airports and terminating airports. ............................................................. 118

Table 5-1: Distribution of No. of Domestic Leg of International and U.S. territory Itineraries

(Data Source: 2007 DB1B Data). ............................................................................................... 124

Table 6-1: Summary of Previous Airport Choice Models in Multi-Airport System. ................. 152

Table A-1: Distance (County – County): Miles. ......................................................................... 192

Table A-2: Distance between County and Airport: Miles. ......................................................... 193

Table A-3: County – County Passenger Demand (Business; Income Group 1). ........................ 194





Table A-8: County – County Passenger Demand (Non-Business; Income Group 1). ................ 199

Table A-9: County – County Passenger Demand (Non-Business; Income Group 2). ................ 200

Table A-10: County – County Passenger Demand (Non-Business; Income Group 3). .............. 201



Table A-13: Time-based Landing Fee. ....................................................................................... 204

Table B-1: 2010 Passenger Traffic for 871 Transatlantic Airport Pairs without Nonstop Service.

..................................................................................................................................................... 216

Table B-2: Enplanements from the CONUS to the Nine International Regions during 1990 –

2008 (Source: 1990 – 2008 T100 International Market Data). .................................................. 222

xxi

Table B-3: Growth of Enplanements from the CONUS to the Nine International Regions during

1990 – 2008 (Source: 1990 – 2008 T100 International Market Data). ...................................... 223

Table B-4: Real GDP of U.S. and the Nine International Regions during 1990 – 2008: Billion

2005 $ (Source: United States Department of Agriculture (USDA) International Macroeconomic

Data Set). .................................................................................................................................... 224

Table B-5: Growth of Real GDP of the U.S. and the Nine International Regions during 1990 –

2008 (Source: United States Department of Agriculture (USDA) International Macroeconomic

Data Set). .................................................................................................................................... 225

Table B-6: Real Per Capita GDP of the U.S. and the Nine International Regions during 1990 –

2008: 2005 $ (Source: United States Department of Agriculture (USDA) International

Macroeconomic Data Set). ......................................................................................................... 226

Table B-7: Growth of Real Per Capita GDP of the U.S. and the Nine International Regions

during 1990 – 2008 (Source: United States Department of Agriculture (USDA) International

Macroeconomic Data Set). ......................................................................................................... 227

Table B-8: Real GDP of U.S. and the Nine International Regions during 2009 – 2040: Billion


Data Set). .................................................................................................................................... 228

Table B-9: Growth of Real GDP of U.S. and the Nine International Regions during 2009 – 2040

(Source: United States Department of Agriculture (USDA) International Macroeconomic Data

Set). ............................................................................................................................................. 229

Table B-10: Real Per Capita GDP of U.S. and Nine International Regions during 2009 – 2040:


Data Set). .................................................................................................................................... 230

Table B-11: Growth of Real Per Capita GDP of U.S. and Nine International Regions during 2009

– 2040 (Source: United States Department of Agriculture (USDA) International Macroeconomic

Data Set). .................................................................................................................................... 231

Table C-1: Passenger Traffic between the United States and the Selected Nine European

Countries in Analysis Domain during 1990 – 2020 (1990 – 2007 Data Source: 1990 – 2007 T100


xxii

Table C-2: Growth of Passenger Traffic between the United States and the Selected Nine

European Countries in Analysis Domain during 1990 – 2020 (1990 – 2007 Data Source: 1990 –

2007 T100 International Market Data). ..................................................................................... 259

Table C-3: Real 2000 GDP of U.S. and nine European countries during 2008 - 2020 (Source:

United States Department of Agriculture (USDA) International Macroeconomic Data Set). ... 260

Table C-4: Growth of Real 2000 GDP of U.S. and nine European countries during 2008 – 2020


Set). ............................................................................................................................................. 260

Table C-5: Passenger Traffic between the United States and the Selected Nine European

Countries in Analysis Domain during 1990 – 2007 (Source: 1990 – 2007 T100 International

Market Data). .............................................................................................................................. 267

Table C-6: Growth of Passenger Traffic between the United States and the Selected Nine

European Countries in Analysis Domain during 1990 – 2007 (Source: 1990 – 2007 T100


Table C-7: Real 2000 GDP of U.S. and Nine European Countries during 1990 – 2007 (Source:

United States Department of Agriculture (USDA) International Macroeconomic Data Set). ... 269

Table C-8: Growth of Real 2000 GDP of U.S. and Nine European Countries during 1990 – 2007


Set). ............................................................................................................................................. 270

Table C-9: Population of U.S. and Nine European Countries during 1990 – 2007 (Source: United

States Department of Agriculture (USDA) International Macroeconomic Data Set). ............... 271

Table C-10: Average Airfare (2000 Year $) from U.S. to Nine European Countries during 1998 –

2007 (Source: 1998 – 2007 DB1B Data). ................................................................................... 272

1

1 Introduction

This dissertation is a collection of several models to understand airline and passenger dynamics

in the National Airspace System (NAS). Specifically, the models are developed to:

a) Simulate the complex interactions between airlines, travelers, and airports of a small-scale

transportation system including two modes of transportation (commercial air and

automobile), ten airports and three airlines,

b) Forecast the international passenger traffic between the CONUS and the rest of the world

(international + non-CONUS, i.e., Hawaii, Alaska and the U.S. territories),

c) Predict the effect of the new Open Skies agreement on the behavior of passengers flying

across the North Atlantic Ocean and airline’s behavior such as opening new nonstop flights,

increase frequency, decrease airfare, and adoption the new available aircraft type based on

the new aircraft performance,

d) Estimate the domestic enplanements generated by the international passengers (DOI

enplanements) at airports within the continental U.S. (CONUS), and the international

passenger traffic through airport-pairs within the CONUS, and

e) Exploit the conditions under which the secondary airports emerge in National Airspace

System (NAS), analyze the markets served at the emerging secondary airports with the

existing network of airport.

1.1 An Agent-Based Model of Airline Evolution


to understanding the dynamic behavior of complex systems. Actions taken by airlines,

passengers, airports, pilots, controllers, the Federal Aviation Administration, and other parties

affect each other.

In Chapter 2, a model is developed to capture the complex interactions between airlines, travelers

and airports, in a small-scale transportation system. The model uses agent-based learning in the

evolution of the airline’s decisions to adjust airfare, adjust aircraft size, open new flights or

cancel unprofitable flights. For a small-scale air transportation system, three airlines, one low

cost and two network airlines are simulated to examine how each airline behaves over time to

2

maximize their profit margins for a given passenger demand and operation cost structure.

Passenger mode choice and itinerary choice sub modules are embedded in the framework to

characterize traveler agent’s response to the evolved airline schedule. An airport delay model

was implemented to estimate the average delay at each airport. The estimated delay fed into the

mode choice and itinerary choice models to update the travel time related variables.

1.2 International Enplanements within the Continental U.S.



increased from 8% in 1990 to 11% in 2008. According to Federal Aviation Administration

(FAA) statistics, international enplanements in the United States grew by 21.4 million between

2000 and 2008 with an annual average growth rate of 4.1% compared with that of 0.7% for

domestic enplanements. The objective of Chapter 3 is to forecast the international passenger

traffic between the CONUS and the rest of the world (international + non-CONUS, i.e., Hawaii,

Alaska and the U.S. territories) between 2009 through 2040.

1.3 The Impact of the EU-US Open Skies Agreement on Commercial Airline Passenger

Traffic over the North Atlantic

The European Union – United States Open Skies Agreement, which became effective March 30,

2008, was established to accomplish several goals including:

• “ … to promote an international aviation system based on competition among airlines in the

marketplace with minimum government interference and regulation.”

• “ … to facilitate the expansion of international air transport opportunities, including through

the development of air transportation networks to meet the needs of passengers and shippers

for convenient air transportation services.”

• “ … to make it possible for airlines to offer the travelling and shipping public competitive

prices and services in open markets.”

The objective of Chapter 4 is to develop mathematical models which permit forecasting the

effect of EU-US Open Skies Agreement on commercial airline passenger traffic over the North

Atlantic Ocean. Specifically, models are developed to predict the passenger traffic amongst the

3

most highly travelled United States to European country pairs, and to predict the traffic amongst

the highly travelled passenger United States airport to European airport pairs. From these

predictions, potential new nonstop flight airport pairs are offered. Lastly, suggestions are made

as to European airports where selected passengers will be rerouted to address the situation when

London, Heathrow Airport reaches full capacity and can accept no additional passenger traffic.

1.4 Domestic Leg of International Passengers within the Continental U.S.

The number of international enplanements within CONUS is predicted to grow from 74.7 million

in 2008 to 184.4 million in 2028. The average annual growth rate is expected to be 4.7%. It was

also showed that the proportion of international enplanements relative to total enplanements

within CONUS increased from 8% in 1990 to 11% in 2008. Some international passengers

generate enplanements not only at the gateway airport, but also domestic enplanements before

(departure trip) or after (arrival trip) the gateway airport. Specifically, 51% of the sampled

international and U.S. territories passengers served by U.S. carriers had at least one domestic

coupon in 2007. The objective of Chapter 5 is to estimate the international passenger traffic

through airport-pairs within the Continental U.S. (CONUS), and to estimate the domestic

enplanements (DOI enplanements) at airports within CONUS generated by the international

passengers during the time period 1990 through 2030.

1.5 Development of Multi-Airport System and Commercial Airline Network in the

National Airspace System (NAS)




objective of Chapter 6 is to understand the conditions under which the secondary airports

develop in a metropolitan area. The markets and connectivity supplied by the secondary airports

are also identified and compared with that of the primary airports. This analysis is needed to

understand and improve the dynamics of the network evolver developed for the Transportation

System Analysis Model (TSAM).

4

2 An Agent-Based Model of Airline Evolution (to be submitted to Journal of Air Transport

Management)

Abstract


to understanding the dynamic behavior of complex systems. Actions taken by airlines,

passengers, airports, pilots, controllers, the Federal Aviation Administration, and other parties

affect each other.

The usefulness of agent-based modeling has been demonstrated by simulating the complex

interactions between airlines, travelers, and airports of a small-scale transportation system. Both

commercial air and automobile modes were simulated. The aviation system covered ten airports

and three airlines. The airline agents show the capability to improve their profit margins by

evolving their market strategies. The airlines’ profit and the associated schedules evolved to a

stable level as expected. The equilibrium profits were improved further when airlines employed

more complex market strategies. The model uses agent-based learning in the evolution of the

airline’s decisions to adjust airfare, adjust aircraft size, open new flights or cancel unprofitable

flights. For a small-scale air transportation system, three airlines, one low cost and two network

airlines are simulated to examine how each airline behaves over time to maximize their profit

margins for a given passenger demand and operation cost structure. Passenger mode choice and

itinerary choice sub modules are embedded in the framework to characterize traveler agent’s

response to the evolved airline schedule. An airport delay model was implemented to estimate

the average delay at each airport. The estimated delay fed into the mode choice and itinerary

choice models to update the travel time related variables.

5

2.1 Introduction

2.1.1 Agent-based model

Agent-based modeling (ABM) is a powerful tool in modeling emergent phenomena [2].

Emergent phenomena is based on the paradigm that the whole system is more than the sum of its

individual entities due to the interactions between the entities. ABM models the complex system

in a more natural way because of its ability to capture emergent phenomena by simulating the

behavior of each entity constructing the system from the bottom up. ABM model have great

flexibility because more agents can be added to the model with time with minimum effort

compared to traditional econometric models, which need to be recalibrated for every new

variable. The modeler can also change levels of description and aggregation by modeling with

aggregate agents, subgroups of agents, and single agents.

In agent-based modeling, a system is modeled as a collection of individual agents representing

the entities constituting the system. The agents can communicate with each other and with the

environment. They produce the overall performance of the system by cooperating, collaborating

or competing, and exchanging information and knowledge at the same time as finishing tasks

independently within the environment. The attributes of the agents are summarized by Kikuchi

[4] as follows:

• Adaptive – The agents can make different decisions as the environment changes.

• Learning – The agents are capable of improving its goal / performance as simulation

runs.

• Autonomous – The agents are able to decide how to act by themselves without human

intervention.

• Goal-driven – The agent’s decision / action are determined by a set of goals. The goals

are assigned to agents based on their attributes.

• Social – The agents cooperate with other agents to produce the system behavior. They

communicate with each other and with the environment.

The main challenge in designing an agent based model is integrating enough knowledge so that

the agents act similar to the real world. The significant computational time required to simulate

such a large-scale ABM system is also a challenge. The ABM develops overall performance of

6

the system by simulating the behavior of the system’s entities from the bottom up. Describing the

individual behavior of all the entities requires a large amount date and extensive computational

time when the system is large.

2.1.2 Motivation

ABM is necessary in order to fully understand large-scale complex systems such as the National

Transportation System [19]. In the National Transportation System, actions taken by various

entities such as airlines, travelers, airports, pilots, controllers, and the Federal Aviation

Administration (FAA) are affected by each other. The complex intersections between airlines,

travelers, and airports are shown in Figure 2-1. The travelers are the revenue source of the airline

and airport. Passenger demand determines the airline’s market strategy. The attributes of the

airlines’ schedule affect the travelers’ decision when choosing their travel mode and itinerary.

Meanwhile, the airport that is a service provider to the airline has to constantly adjust to meet

airlines operational demand requirements. Too many flight operations result in congestion at the

airport due to insufficient capacity. The level of delay affects airlines and passengers’ behavior

to choose the transfer airports. Traditional economic modeling tools are not able to model such

complex interactions that require multi-feedback loops [19]. ABM on the other hand is

developed to analyze such systems.

Figure 2-1: Interactions between Airlines, Airports and Travelers in the National Transportation

System.

7

2.1.3 Objective

The ABM aims to capture the complex interactions between airlines, travelers and airports, in a

small-scale transportation system. The small-scale transportation system we model in this

research effort is shown in Figure 2-2. It includes fifteen counties and ten airports with three

airlines providing air service between the airports as shown in the figure. The distance between

counties and the distance between the county and airport are presented in Appendix A.4. Each

participant in the system is modeled as an individual agent that has its own ability and volition to

take actions so that the objective function is maximized. The airline agents adjust their strategies

to maximize the profit through agent-based learning. The traveler and airport agents respond to

the airline agent behavior and at the same time change the environment of the transportation

system (passenger demand, delay). The airline agents adjust their actions as the environment

changes.

Figure 2-2: Small-scale Transportation System.

2.2 Literature Review

A summary of previous ABM models that have been applied to simulate entities in the National

Transportation System (NTS), or its subset the National Airspace System (NAS) are shown in

Table 2-1.

7

29

38

10

1

4 5

6

3

4

13

9

14

8

7

15

12

6

1

2

11

5

10

1500 Mile

Airline AAirline BAirline C

County CentroidBig Hub AirportMedium Hub AirportSmall Hub AirportNonHub Airport

8

Table 2-1: Summary of agent-based models simulating the key participants in the National

Transportation System (NTS).

Agent-Based Model

Langerman’s dissertation

Kim’s dissertation

Jet:Wise TransNet

Geographical Scope

Hypothetical Air Traffic Network

Hypothetical Air Traffic Network

CONUS CONUS

Travel Mode Airline Airline Airline Auto, CA, GA

Agent Type Airline, Airport, Aircraft

Airline Airport, Traveler

Airline, Flight, Passengers, Airport

Airline, Traveler

Nodes 3 Airports 10 Airports 191 (MAs) 4 locales Competing Airlines

1 3 All the real airlines in NAS

3

Aircraft Types 3 3 All the real aircraft types in NAS

Unknown

O-D Markets 6 Airport pairs 90 (Airport pairs)

2,493 (MA pairs) Unknown

Flight Routes 6 172 74,471 Unknown Flight Legs 6 36 Unknown Unknown Generation Run 23 400 1,400 Unknown

Langermann [6] presented an agent-based model to solve the airline scheduling problem and re-

optimize the airline scheduling after a disruption occurs. In this model, an organization consists

of aircraft agents, airport agents, a business manager agent and a resource manager agent to

simulate the airline operating environment.

Kim [5] developed an agent-based model to analyze air traffic congestion and airline competition

using the fuzzy logic system in his dissertation work. The behavior airlines, airports and

passengers are simulated as agents in the model. The complex interplays among agents and the

resulting emergent phenomena are analyzed.

Transportation Network (TransNet) [8, 9] is another model that simulates the interactions

between the consumer agent and each service provider agent in the National Transportation

System (NTS). The model aims to capture the dynamic interaction between the consumer mode

choice and route choice behavior with the service provider at the microscopic level. Langermann

[6] presented an agent-based model to solve the airline scheduling problem and to re-optimize

9

the airline scheduling after a disruption occurs. In this model, an organization consists of aircraft

agents, airport agents, a business manager agent and a resource manager agent to simulate the

airline operating environment.

Jet:Wise [11, 12, 13] is an agent-based model developed at The MITRE Corporation’s Center for

Advanced Aviation System Development (CAASD) to simulate complex, emergent behavior

observed in the U.S. National Airspace System (NAS). Jet:Wise models the evolution of the

airline industry to assign which aircraft to fly, where and when to serve an airport, where to

establish hubs, what business and leisure fares and airline response to delays, congestion and

missed connections. The model also simulates the departure/arrival restriction decisions made by

the Federal Aviation Administration (FAA) and airports. Jet:Wise models the continental United

States (CONUS) composed of 191 Metropolitan areas. Jet:Wise applies 13 learning tools,

including 7 continuous and 6 discontinuous, to simulate the evolution of each airline’s strategies.

Airline agents learn to maximize their profit margins by applying the learning tools for hundreds

or thousands of iterations. In each of the iterations, only one learning tool is applied to all of the

airlines.

Agent-based modeling techniques have been used to model the Air Traffic Management system

[9, 10, 14-22]. An example is the Intelligent agent-based Model for Policy Assessment of

Collaborative Traffic flow management (IMPACT) [20] is such a model developed by MITRE-

CAASD. The model aims to simulate the complex decision-making interactions between airlines

and ATM authority when weather disrupts the airline schedule. In this model, ten to fifteen

airline agents and a single FAA agent are modeled as self-interested decision makers in the

traffic flow management (TFM) system. Airport Demand/Capacity Model (ADCM) [21] is an

agent-based model for operational analysis of airport policy changes. The model provides a

prediction of airport demand changes considering the interdependency of all the key users at the

airport. Optimal Aircraft Sequencing using Intelligent Scheduling (OASIS) [9] is another

application of agent-based techniques in air traffic management. The air traffic management

system was modeled by a set of global agents consisting of a Coordinator agent, a Sequencer

agent, a Trajectory Checker agent, a Wind Model agent, a User Interface agent and aircraft

agents. The agents cooperate to maximize runway utilization. Ryan [14] proposed an agent-based

model to simulate the air traffic activities within one sector. The model was used to examine the

10

efficiency of imposing a route fee to increase the capacity of the sector. The airlines, flights and

the controller are modeled as autonomous agents. Each agent makes decisions to meet its own

goal. The airline agents decide routes to maintain an economic advantage. The flight agent is

modeled to make a response when necessary. The controller agent aims to maintain the safety

(separation) within its sector.

In addition, ABM modeling techniques also have been applied to travel demand forecasting.

Zhang and Levinson [23] developed a model to solve complex transportation problems of trip

distribution and route assignment by applying simple local rules of agent behavior. In this model,

the macroscopic travel demand is modeled as emergent phenomena from the interactions

between microscopic agent behaviors in the transportation system. Three types of agents: node,

arc and traveler, are used to model the transportation network. Instead of using discrete choice

analysis, the travel agent is capable of identifying the shortest route through learning from other

travelers and from its own experience. The traveler agent is motivated to find a travel activity

and to reach the activity destination with the lowest travel costs. The node agent learns the

shortest paths from other nodes to itself from travelers, and then distributes the information to

travelers when its knowledge is superior to the travelers. The arc agent obtains its arc cost based

on the arc capacities and assigns the cost to the travelers. Bonabeau [2] provides example

application of agent-based modeling in business context: flows, markets, organizations and

diffusion.

2.3 Methodology

2.3.1 Overview of the ABM Model

The model developed for this research integrates travelers’ mode choice, commercial air

passengers’ itinerary choice, and projected airport delays. Two modes of transportation,

automobile and commercial air, are considered in the mode choice. All itineraries connecting the

origin and destination airports are considered as alternative itineraries. The model inputs are the

deterministic daily passenger demand between counties, three distance matrices (airport to

airport, county to county and airport to county), the capacity of each airport, the airline network,

travel time, travel cost related variables. All input data is included in the Appendix A.4. The

outputs of simulation include the evolved airline schedule, and associated market share of

11

commercial air and airline performance statistics including profit, load factor, market share and

other variables.

Several of scenarios examined are representations of conditions that the airlines might face in the

future. For each scenario, the model is run for a predefined number of generations. The airline

agents can then maximize their profit margins by evolving their strategies. For the evolution step

in generation zero, the air schedule, itinerary, airfare and the airline strategy, are initialized for

each airline (Figure 2-3). In subsequent generation steps, the airlines evolve their strategies by

going through the following steps:

1. The airlines update their schedules, airfares and itineraries based on the strategy applied

in the current generation. For example, when the airlines adjust aircraft size, only

schedules need to be updated. Both schedules and itineraries need to be updated when

airlines cancel or add new flights.

2. The updated airline schedules and the assumed capacity of each airport are inputs to the

airport delay model to estimate the average delay per operation.

3. Given the average delay of each airport, attributes of automobile and airline schedules,

the mode choice model splits a given passenger demand between counties by automobile

and commercial air, and then estimates the air passengers traveling between airports.

4. The itinerary model estimates the passenger demand for each flight by assigning the air

passengers among the alternative itineraries.

5. For each generation in the simulation, airline performance statistics such as profit, load

factor and market share are estimated based on the demand for flight. The airlines

compare these statistics with the corresponding values in the previous generation to

evolve the market strategy.

6. Repeat if the current generation is not the last generation, or output all the statistics and

end the simulation.

12

Figure 2-3: Flow Chart of Airline Strategy Evolution Process.

3. Travelers Mode Choice Model

1. Update Evolved Parameters (i.e. Airline Schedule, Itineraries, Air Fare)

N

Start

n = 0Initialize Airline Schedule, Itineraries & Airfare

Initialize Airline Strategy

6. Evaluate the Objective Function& Evolve the Airline Strategy

n = n + 1n = 1 ?

N

Y

End

n > N* ?

Output the Statistics

Y

Iteration

2. Airport Delay Model

4. Air Pax. Itinerary Choice Model

5. Calculate the Airline Performance Statistics

* N - Predefined Simulation Length

13

2.3.2 Traveler Agent Behavior

Traveler agents are assumed to maximize their utilities obtained from alternative modes of

transportation and itineraries when they choose their travel mode and itinerary respectively. The

travelers’ behavior of choosing the mode of transportation and itinerary are simulated using the

logit model as shown in Equation 1.

∑ Equation 2-1

where is the probability that decision maker n chooses alternative i. is the representative

utility associated with alternative . The representative utility is usually specified to be a linear

function of the explanatory variables with the form in Equation 2 [Train]. The explanatory

variables are usually the attributes of the alternatives and attributes of decision makers.

∑ Equation 2-2

where is explanatory variable that relates to alternative as faced by decision maker .

is the coefficient of attribute variable . The coefficients show the relative importance that

decision makers place on the explanatory variables. The coefficients are constant when all the

decision makers are assumed to value all the explanatory variables with the same perception.

When taste variation exists due to unobserved reasons affecting people’s decision, the.mixed

logit model with random coefficients of explanatory variables should be employed. The utility

function form of mixed logit model is shown in Equation 3:

∑ Equation 2-3

where is the fixed coefficient of the explanatory variable j. is a random coefficient with

zero mean. The normal distribution is usually imposed on , but it should be noted that normal

distribution can produce counterintuitive probabilities because it ranges over the positive and

negative scale. Given the utility function form, the coefficients of attribute variables are

calibrated from the observed data. Reasonable coefficients should be both statistically significant

(with reasonable -value) and intuitive with regard to the sign. For example, the coefficient of

airfare variable in one itinerary’s utility function is expected to be negative. This is because the

itinerary with the lowest airfare has the highest probability to be selected among all the

alternative itineraries when all the other attributes are same.

14

2.3.2.1 Traveler Mode choice model

The mode choice model uses a nested logit model to estimate the choice between automobile and

commercial air mode in an upper nest. A lower nest under the commercial air choice is used to

estimate the choice between the candidate origin and destination airports. There are up to a

maximum of nine routes between the three origin and three destination airports. The inputs of the

mode choice model include the assumed daily passenger demand between counties by five

income groups and two trip purposes, and travel time and travel cost related variables for both

automobile and commercial air (all inputs are included in the Appendix A.4). Note that the

deterministic average daily passenger demand is used in this model. That is, the demand

variation due to season of year, day of week etc. is not considered. The output of the mode

choice are probabilities of choosing between airports and the aggregate market share of

commercial air.

The market share of each transportation mode is estimated. The demand by mode between

county pairs is obtained by multiplying the demand by the respective market share of each travel

mode. The air passenger demand between county pairs is further split among routes using an

airport choice model. The airport choice model locates three airports closest to each county’s

centroid and assigns them as the county’s candidate airports.

Travel time and travel cost are considered as the explanatory variables in the travelers’ mode

choice. The utility for each alternative mode is estimated with Equation 4:

Equation 2-4

where is the utility that travelers in income group obtain from transportation mode from

origin county to destination county { 1, 2 (1 for auto; 2 for commercial air); 1 to 5;

, 1 to 15}. is the travel time coefficient. , , , , are the travel cost coefficients

for income group 1 to 5 respectively. The coefficients are calibrated from American Travel

Survey (ATS) 1995 data [BTS]. The coefficients are calibrated separately for business travelers

and leisure travelers. More details about the travelers’ mode choice are presented in Ashiabor et

al. [1].

15

2.3.2.2 Commercial Air Passenger Itinerary Choice Model

Itinerary level forecasts on are important for accurate airline planning and flight schedule

construction since itineraries are the products purchased by passengers [24 - 26, 28]. The

commercial airline itinerary choice model allocates the passenger demand between airports to

various itineraries based on their market shares. An itinerary is defined as either a direct flight or

a sequence of flights connecting a specific origin and destination airport pair. In the model, all

itineraries linking the specific origin and destination airport pair are considered as alternative

itineraries, although in some itineraries big time gaps may exist between passengers’ desired

departure times and the actual departure times.

The representative utility of each alternative itinerary is assumed to be a linear function of the

itinerary attribute variables including schedule delay, total travel time, airfare and the number of

stops. The schedule delay of an itinerary is defined as the time gap between passengers’ desired

departure time and the actual departure time of the itinerary. The departure time of an itinerary is

defined as the departure time of the first flight included in the itinerary. In this model, both

passengers’ desired departure time and flights’ departure time are specified by a time window of

fifteen minutes. The itineraries’ maximum number of stops is two since an itinerary is assumed

to include at most three flights.

Equation 2-5

where represents the utility of itinerary for a passenger who desires to depart in time

window from airport to airport ( , 1 to 10 & ; 1, 2, 3, …; 1 to 96).

is the schedule delay of itinerary for a passenger who desires to depart in

time window . , , , are the coefficients for attribute variable Travel Time, Airfare,

Schedule Delay and Number of Stops respectively. The survey capturing the passengers’

revealed preference or stated preference, or the airline supplied booking data is usually used to

calibrate the itinerary choice coefficients [24 - 26, 28].

In this model, a day is split into ninety-six time windows of fifteen minutes. The daily passenger

demand between airports is distributed among time windows based on the assumed distribution

of passenger desired departure time. Ninety origin and destination airport pairs exist in the small-

16

scale air transportation system including ten airports. All passengers are classified into 8,640

(90x96) groups by their desired departure time and origin and destination airport pair.

Figure 2-4: Flowchart for Itinerary Choice Model.

17

The commercial airline passengers’ itinerary choice is more complex than simply allocating all

of the passengers from a demand group to the various alternative itineraries at one time.

“Interference” exists between the demand groups. The passengers from each demand group may

compete for the same itinerary. The passenger demand for an itinerary is constrained by the

itinerary capacity. In reality, the state of airline reservation system evolves over time. An

itinerary cannot accept passengers when any flight included in the itinerary is full.

The flow chart of the itinerary choice model is shown in Figure 2-4. The commercial airline

passenger itinerary choice behavior is simulated using the following procedure:

1. Randomly select a passenger demand group using the uniform distribution over all the

groups.

2. Determine all alternative itineraries for the chosen passenger demand group.

3. Allocate a small number of passengers from the chosen group among all alternative

itineraries using the logit model (the smaller the number of passengers allocated at a time,

the more accurate the simulation). Passengers rejected by one itinerary are then assigned

to the rest of the alternative itineraries.

4. Update the alternative itineraries and their capacities.

This procedure is applied repeatedly until all of the passengers are either assigned to the

itineraries or deemed to be missing passengers when no alternative itinerary is available.

18

2.3.3 Airline Agent Behavior

Figure 2-5: Hypothetic Airline Network and Distance (Miles) Between Airports.

Three airlines are simulated in this model. One low cost airline (A) and two network airlines (B

and C). The characteristics of these airlines and their network are given in Table 1 and Figure 2-5

respectively. The numbers on the plots in Figure 2-5 represent the distance between airports.

Table 2-2: Airline Characteristics.

Airline A B C

Network As shown in Figure 2-5

Total Flights (at the beginning) 116 120 150

Airfare above Average -20% 0 20%

Operation Cost Multiplier 3 4 5

2.3.3.1 Schedule Generator

The schedule generator builds the initial airline schedules for the small-scale air transportation

network in generation zero. The frequency is randomly selected from the assumed frequency

ranges between three and seven for all legs. All flights are randomly allocated into ninety-six 15

minutes time windows based on the assumed distribution of passengers’ desired departure time.

Only one flight is allowed per time window.

7

2

9

38

10

1

4 5

6

Airline AAirline BAirline C

7

2

9

38

10

1

4 5

6660

830

280340

510

700220

250

230

320390

250

400

400 700270

800630

19

The initial aircraft type for each flight is randomly selected from the three given aircraft types.

The characteristics of these aircraft types are shown in Table 3.

Table 2-3: Aircraft Characteristics.

Aircraft Type Boeing 757 Boeing 767 Boeing 777

Fuel Consumed (Gallon / 15 Min.) 550 710 860

Weight (lbs) 220,000 345,000 506,000

Landing Fee ($) 1,540 2,415 3,542

Cruise Speed (mph) 530 530 560

Seat capacity (1 Class) 186 260 320

2.3.3.2 Itinerary generator

The itinerary generator builds all alternative itineraries for the specific origin and destination

airport pair based on the flight-based airline schedules. Code-sharing is not modeled in this

small-scale air transportation system. All flights included in a connecting itinerary belong to the

same airline. In reality, one airline may sign a code-sharing agreement with another airline so

that they can make use of each other’s capacity. All of the top 20 airlines in the world share their

route structure with at least one other airline to gain higher market share [30].

The nonstop itineraries are straightforward to generate because only one flight is involved. It is

more complex to construct the connecting itinerary including multiple flights. Starting from the

nonstop itineraries, an algorithm is adopted to generate the itineraries with ( > 1) connections

from the itineraries with ( - 1) connections. The following procedure is used to generate the

alternative itineraries for the specific origin and destination airport pair.

For 1

1. Initialize

2. For each found from the airline schedule list, construct a row

vector ]

3. Update by inserting the new row vector:

; ] }

For l 2, 3, …

20

4. Initialize Itinerary

5. For any nonempty Itinerary and Flight , add m t t ] at end of each

row vector of Itinerary to construct a new itinerary matrix as

Itinerary _New

6. Update Itinerary by inserting the new itinerary matrix:

7. Itinerary Itinerary ; Itinerary _New

where represent airline ’s alternative itineraries connecting origin airport and

destination airport with the number of flights ( , 1 to 10, ; 1, 2, 3; 1, 2, 3,

…). It is a matrix when several itineraries containing flights exist for the given origin and

destination airport pair. represents the unique flight originating from airport and

destining to airport with departure time and arrival time ( , = 1 to 96).

The alternative itinerary also needs to meet minimum and maximum connection time

requirements at the connecting airport. The minimum value usually is the shortest time in which

passengers embark from one plane and board another. The maximum value is the longest time

that passengers are willing to wait for the connecting flight. The shortest and longest connection

times may differ between airports of different hub types. Forty-five minutes and two hours are

assumed as the minimum and maximum connection time respectively for all airports in the

model.

The airfare of low cost airline A is assumed to be twenty percent lower than the average airfare

level. The airfares of network airlines B and C are assumed to be at average level and twenty

percent higher than the average level respectively. The initial average itinerary airfare is

estimated in generation zero by Harris model with the form in Equation 6.

Equation 2-6

where represents the average airfare of an itinerary connecting airport and airport .

The superscripts , are the hub type of origin airport and the ticket class respectively. Four

hub types (large hub, medium hub, small hub and non-hub) and two ticket classes (business class

and coach class) are used in the model. is the total flown distance of all

21

flights included in the itinerary. , , are calibrated constant coefficients

using the itinerary data from the 10% ticket sample. The coefficients are calibrated separately by

four origin airport’s hub types and two ticket classes.

The profit of each flight is estimated by the flight revenue minus the operation cost. The airline

revenue in this model is assumed to come completely from airfare revenue. In reality,

miscellaneous sources like cargo and advertisements also contribute to airline revenues.

A flight might serve both direct passengers and connecting passengers. The nonstop itinerary

includes one single flight. The itinerary fare of direct passengers goes to the revenue of this flight

directly. Since connecting itineraries includes more than one flight, itinerary fares of connecting

passengers need to be split among the flights included in their itineraries. The nonstop itinerary

fare that direct passengers pay for each flight is used as a weight factor to split the connecting

itinerary fare among flights.

The total operation cost of each flight in this model is estimated by multiplying the landing fee

and fuel cost with an operation cost multiplier as shown in Equation 7.

Equation 2-7

Where represents the total operation cost of flight belonging to airline . As

shown in Table 2, is 3, 4 and 5 for airline A, B and C respectively.

The operation cost multiplier is used for two purposes. First, it reflects the existence of other

operation costs other than the landing fee and fuel cost. The operation cost usually consists of

direct operation cost and indirect operation cost. The direct operation cost usually refers to fees

related to the flight operations such as fuel cost, the landing fee and the maintenance fee. The

indirect operation cost usually derives from passenger service, aircraft service, ground property

and other administrative concerns. Second, different cost structure between airlines is modeled

by assigning different values of operation cost multiplier to airlines. The total operation cost is

assumed to be three, four and five times the fuel cost and landing fee for airline A, B and C

respectively. The widely used weight-based landing fee of $7 / 1,000 lbs is assumed for all

aircraft types in this model.

22

2.3.3.3 Airline Strategies and evolution rules

The airline agents are assumed to maximize their profit margins by evolving four strategies:

adjust the airfare of each flight, adjust the seat capacity of each flight, add flights to the segment

with potential profit and cancel the unprofitable flights. Airline agents evolve these strategies in

a predefined order. Each strategy is applied by the predefined frequency. Airlines are assumed to

use the same order and frequency. As shown in Table 4, in every iteration airline agents adjust

airfare for the first ten generations, add flights for one generation, adjust the aircraft size for ten

consecutive generations, and finally cancel flights for one generation.

Table 2-4: The Predefined Order and Frequency Airline Agents Use to Evolve the Strategies.

Iteration 1 2 3 - 50

Strategy \ Generation 1 - 10 11 12 - 21 22 23 - 32 33 34 - 43 44 - - - -

1 – Adjust airfare √ √ √

2 – Add flights √ √ √

3 - Adjust aircraft size √ √ √

4 – Cancel flights √ √ √

Airline agents evolve strategies by controlling the evolution direction and the evolution step. The

evolution direction determines whether to increase or decrease the related schedule variable.

Evolution step determines how much to increase or decrease. At the beginning of each

generation, airlines evolve certain strategies with certain direction and step. The resulting airline

profit is checked at the end of each generation. A strategy evolution reinforces the airline profit

when the strategy evolution improves the airline profit. The reinforcement times represent

consecutive airline profit improvements due to one strategy evolution. The reinforcement times

can be negative. A negative number of reinforcement times indicates the number of consecutive

decreases of airline profit caused by certain strategy evolution. Based on the evolution direction,

step and the reinforcement times in each generation, airline agents apply the following rules to

evolve the strategies. One example is included in Table 5 to show how these rules are used to

apply the strategy of adding flights.

23

Table 2-5: An Example to Show the Rules Used by Airline Agents to Evolve Strategies.

Iteration 1 2 3 4 5 6 7 8 9 10 11 - 49 50 Generation 11 33 55 77 99 121 165 187 209 231 - 1089 No. of flights to add 1 1 1 2 2 1 1 0 0 0 - 1 Reinforcement times +1 +2 +

3 +1 -1 -2 -3 0 0 0 - +2

Airline profit (↑- Increased; ↓- Decreased)

↑ ↑ ↑ ↑ ↓ ↓ ↓ Skip 1

Skip 2

Skip 3 - ↑

The strategy used in the example is to add flights. The order and frequency to apply such

strategies are same as Table 2-4. The ‘Generation’ row in this table lists all the generations when

airlines have the chances to add flights.

Keep the previous evolution step: Repeat the direction and step in the next generation to

evolve the same strategy when the strategy evolution in the current generation improves the

airline profit. In the example, one flight is added in generation 33 when the airline profit is

increased in generation 11 with adding one flight.

Double the previous evolution step: Double the evolution step of one strategy when this

strategy evolution improves the airline profit n times consecutively with the same step.

Since consecutive successful applications of the strategy in one direction may imply a

profitable trend at that direction, doubling the learning step is helpful to increase the

evolution speed. n = 3 is used in the example. Adding one flight within the first three

iterations all increased the airline’s profit, so the airline adds two flights in the iteration 4

when the airline has the chance to add flights again.

Change the evolution direction and reduce the evolution step: Reduce the evolution step by

a predefined ratio for the strategies with continuous step, or reduce it to one for strategies

with discrete step. Meanwhile change the evolution direction when the application of this

strategy is not profitable. In the example, only one flight is added in iteration 6 because

adding two flights in iteration 5 decreased the airline’s profit.

Decrease the frequency to apply the strategy: Skip next m generations to apply one strategy

if the airline profit decreases consecutively in the previous m generations when this strategy

is applied (reinforcement times = -m). m = 3 is used in the example. The airline skipped the

24

next three chances to add flights in iteration 8, 9 and 10 after the profit decreased in iteration

5, 6 and 7 when the airline added flights.

To maximize the profit margin, airline agents increase airfare of each flight, but this strategy

must be balanced with demand. Airfare plays an important role in traveler’s mode choice and

itinerary choice. Airfare directly affects passenger demand. When airlines charge too high an

airfare for one itinerary comparing with their competitors, they may lose money on this itinerary

due to low passenger demand. On the contrary, if airlines charge too low an airfare, they may not

make money or their profit margins may be too low despite the fact that their flights are full.

Trade off between airfare and demand must be carefully managed for each flight to maximize

airline profit margins.

Airline agents in the model have to manage their aircraft size for each flight so that their profit

margins can be maximized. Airline agents may improve their profit margins by assigning smaller

aircraft to the flights with low demand and by assigning larger aircraft to the flights with high

demand. The flight profit depends on seat capacity. Both revenue and operation cost of a flight

are a function of its seat capacity. The number of passengers that flight can take is constrained by

its seat capacity. By assigning larger aircraft to high demand flights airlines can potentially

increase profit margins because the increased revenue usually overshadow the increased

operation cost of a large aircraft. Conversely, airline agents may still increase profit margins

when they assign smaller aircraft to low demand flights as the decreased operation cost offsets

the reduced revenue. The fleet mix of all airlines is assumed to satisfy all of the aircraft size

need. This assumption allows the airline agents to search the optimum seat capacity of each

flight by the continuous step. The operation cost of the flights with evolved seat capacity is

obtained through interpolating the operation cost and seat capacity of three given aircraft types.

Airline agents add flights to one or several segments based on the spilled passengers of each

segment. Spilled passengers are those who are rejected by a flight due to an insufficient number

of seats. Only one flight addition is allowed on each segment in one generation. For example, if

four flights are evolved to add in a generation, then one flight is added on each of the four

segments with the most spilled passengers. As shown in Equation 8, spilled passengers are

estimated by comparing the flight’s unstrained passenger demand with its seat capacity.

25

0, Equation 2-8

where is the number of passengers rejected by flight .

is the actual seat capacity of flight . represents the

unconstrained passenger demand for flight due to all itineraries including the flight as one of

the legs. As shown in Equation 9, the unconstrained passenger demand for each itinerary is

estimated from the itinerary choice model when the seat capacity of all flights are assumed as

infinite.

∑ ∑

Equation 2-9

where is the passenger demand from airport to airport with desired

departure time . is the utility of itinerary for a passenger who desires to depart in time

window from airport to airport . It is estimated with Equation 5.

Airline agents cancel one or several flights with the greatest loss. Note new flights may not be

profitable before applying any strategy to adjust its airfare and aircraft size. It is more reasonable

to cancel the unprofitable new flights after making sure that the airlines have already applied

other strategies to these flights. Canceling new flights shortly after they are added is avoided in

this model because airline agents always apply the strategy of adjusting aircraft size in ten

consecutive generations after adding flights.

2.3.4 Airport Agent Behavior

The effect of airport delay on the passengers’ mode choice and itinerary choice is considered in

this model. Congestion exists at the hub airport when the operation demand exceeds the airport

capacity at a particular peak hour. The airport delay affects the airline’s performance as well as

the passengers’ mode choice and itinerary choice. The airline schedules determine the operation

demand of each airport. Based on the airline schedules and the assumed airport hourly capacity,

the average delay per operation by time of day at each airport is estimated using the deterministic

queuing theory. The estimated delay is then input to both the mode choice model and itinerary

choice model to update the travel time related variables.

26

Further action can be assigned to airport agents in efforts to alleviate congestion by evolving the airport demand management

strategies. This will be addressed in more detail in the recommendation section.

All main variables used in the analysis and their intersections are summarized in Figure 2-6.

Figure 2-6: Main Variables and Their Intersections.

27

2.4 Simulation Results

The model was run using three scenarios. Each scenario was run for hundreds of generations.

One generation takes around two minutes to run. In the first scenario, the fuel price is assumed to

be $1.5/gal and the airlines apply two strategies to adjust the airfare and aircraft size of each

flight. In the second scenario, the higher fuel cost of $2.0/gal is imposed on the airlines while

keeping the same strategies. Recall Equation 7, the operation cost of each flight is a function of

the fuel cost. Raising fuel price reflects the situations when airlines impose a higher operation

cost. In the third scenario, as compared with the first scenario, two more strategies are used by

airlines to cancel unprofitable flights and to add flights with high profit potential while keeping

the same fuel price of $1.5/gal.

The evolutions of all airline schedules and the performance metrics of all three scenarios are

shown in Figure 2-12. The results show that all three airlines are profitable at equilibrium profit

for all scenarios. The aggregate market share of commercial air, the average load factor, the

average aircraft size and airfare all reached the equilibrium state. The profit margins of all three

airlines improved as the simulations proceeded. The improvement was fast at the beginning of

simulations and gradually declined by the end of simulations.

In all figures shown in this section, blue plots represent low cost airline A. Green plots and red

plots represent network airlines B and C, respectively. The x-axle represents the generation

index.

The results from the 1st scenario are compared with those from the 2nd scenario in Figure 2-7,

Figure 2-8 and Figure 2-9. As would be expected, the equilibrium profit margins and average

airfares of network airlines (red & green plot) dropped slightly when the operation cost was

increased. It is interesting to observe that the low cost airline improved its equilibrium profit by

decreasing airfare. As shown in Figure 2-8, commercial air gained more market share when the

operation cost was increased. The market share of commercial air is a function of both fuel price

(automobile travel cost related) and airfare. The decrease of average airfare shown in Figure 2-9

may explain the higher market share of commercial air and higher load factors. It is observed

from Figure 2-9 that the higher operation cost had little effect on the airlines’ evolution of

aircraft size since the equilibrium seat capacity of all three airlines was almost the same before

and after the fuel price was increased. The increasing operation cost affected the low cost

28

airline’s (lowest airfare in blue plot) airfare evolution the most in comparison to the equilibrium

airfares, while the higher operation cost almost did not affect the network airlines’ (green and red

plot) airfare evolution.

Figure 2-7: Comparison of Airline Profit & Load Factor between 1st and 2nd Scenario

Figure 2-8: Comparison of Market Share of Airlines & Aggregate Market Share of Commercial

Air between 1st and 2nd Scenario.

29

Figure 2-9: Comparison of Evolutions of Airline Average Airfare & Seat capacity per Flight

between 1st and 2nd Scenario.

Results from the 3rd scenario including evolving airfare, aircraft size and canceling / adding

flights are shown in Figure 2-10. Airline profit margins, average load factors and other airline

performance metrics, along with all of the associated airline schedule variables, converged to

stable level for all airlines.

Airline profit margins and the average load factor from the 1st scenario are compared with those

from the 3rd scenario in Figure 2-11. As expected, airlines improve their equilibrium profit

margins and load factors by applying additional strategies to cancel unprofitable flights and by

adding flights with high profit potential. The equilibrium airfares decreased and the airfare

difference between airlines diminished. Both observations might be explained by the higher

competition between the airlines who gain more freedom to control their profit margins using

more complex strategies.

30

Figure 2-10: Airline Profit, Load Factor, Airfare and Seat capacity of 3rd Scenario with Evolving

Airfare, Aircraft Size and Canceling / Adding Flights.

Figure 2-11: Comparison of Airline Equilibrium Profit & Load Factor between 1st (right plots)

and 3rd (left plots) Scenario.

31

Figure 2-12: Evolution of Airline Schedule and Performance Statistics (Left plots: 1st Scenario*; Middle plots: 2nd Scenario*; Right plots: 3rd Scenario*).

* 1st Scenario: Adjust Airfare and Aircraft size with Fuel Price: $1.5/gal; *2nd Scenario: Adjust Airfare and Aircraft size with Fuel Price: $2.0/gal;

*3rd Scenario: Adjust Airfare, Aircraft size and Add/Cancel Flights with Fuel Price: $1.5/gal;

32

2.5 Conclusions and Recommendations

The research has demonstrated the ability of agent-based models to simulate the emergent

behavior of a small-scale transportation system consisting of two modes of transportation, ten

airports, three airlines and passenger demand between fifteen counties. The airline agents

improve their profit margins by evolving their market strategies. As expected, the airlines’ profit

margins and other performance metrics converged to an equilibrium level. The equilibrium profit

margins were improved when airlines employed more complex market strategies.

The model may help FAA examine the efficiency of the time-based landing fee policy in efforts

to keep airport delays within reason if the airport agents are allowed to charge airlines more for a

flight to land during peak periods. Demand exceeds capacity at many big hub airports where

weight-based airport landing fee policy adopted currently in the model. The uniform congestion

fee at a particular time regardless of the aircraft size or type is expected to reduce the congestion

bias caused by the current weight-based landing fees system [3, 27, 29]. One such landing fee

example is shown in Appendix A.4. The congestion fee can evolve over time to align the demand

to capacity when it is set too low or too high initially. In addition to time-based landing fee

policy, if the airline agent is allowed to adjust the departure time / arrival time of each flight, the

airlines may move some of their flights to an off-peak time so that the demand can be smoothed

over the day, therefore the congestion is alleviated.

Additional strategies may be tested by airline agents to improve their objective function. In the

current model, departure time of each flight is fixed after it is generated in the generation zero.

As mentioned early in this section, it might be necessary for airline agents to reschedule flights

to avoid delays and associated congestion fees. Additionally, considering flight departure time is

closely related to passenger’s schedule delays and total travel time (connecting itinerary), the

airline may reschedule the departure time of each flight to reduce the passengers’ schedule delay,

thus attracting more passengers. The arrival time of each flight may be adjusted to allow more

convenient connections. Multi-class airfare may also be adopted to reflect the practical operation

more closely. One class ticket is assumed in the model for all flights. Under practical yield

management, airlines usually employ a multi-class ticket system to meet an array of passenger’s

needs so that their profit margins can be maximized.

33

The itinerary choice model should be applied separately for business and leisure passengers

considering that business passengers are more sensitive to departure / arrival time and travel

duration, while leisure passengers are relatively more price sensitive. To model that leisure

passengers usually book their tickets earlier to receive discounted airfare, the itinerary choice

model may be applied to the leisure passengers by leaving a certain proportion of seats for

business class passengers.

2.6 References

[1] Ashiabor, S., Baik, H. and Trani, A., “Logit Models to Forecast Nationwide Intercity Travel Demand in the United States,” The 86th Annual Meeting of Transportation Research Board, Washington, D.C., 2007

[2] Bonabeau, E., “Agent-based modeling: Methods and techniques for simulating human system,”, PNAS, vol. 99, suppl. 3, pp. 7280-7287, 2002

[3] Evans, A. D. and Clarke, J-P, B., Responses to airport delays - a system study of Newark International Airport. Report, MIT International Center for Air Transportation, Department of Aeronautics, Massachusetts Institute of Technology, Cambridge, MA, 2002

[4] Kikuchi, S, Rhee, J. and Teodorović, D., “Applicability of an Agent-based Modeling Concept to Modeling of Transportation Phenomena,”, Yugoslav Journal of Operations Research, no. 12, pp. 141-156, 2002

[5] Kim, J., An Agent-based Model for Airline Evolution, Competition and Airport Congestion. PHD Dissertation, Civil and Environmental Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA, 2005

[6] Langerman, J. J., Agent-based models for the creation and management of airline schedules. PHD thesis, Computer Science, University of Johannesburg, 2005

[7] Lewe, J., An Integrated Decision-Making Framework for Transportation Architectures: Application to Aviation Systems Design. PHD thesis, School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, GA, 2005

[8] Lim, C., An integrative assessment of the commercial air transportation system via adaptive agents. PHD thesis Proposal, School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, GA, 2006

[9] Ljungberg, M. and Lucas, A., “The OASIS Air Traffic Management System”, Proceedings of the Second Pacific Rim International Conference on Artificial Intelligence, PRICAI’92, Seoul, Korea

[10] Mehta, S., Nedelescu, L., Nolan, M., Krull, K., Whitford., J., Pfleiderer, M., Foong, C., “Using Agent-Based Simulation to Evaluate Technology and Concepts for the National Airspace System,” IEEEAC, paper # 1227, 2005

[11] Niedringhaus, W., “An agent-based model of the airline industry,”, The MITRE Corporation, McLean, VA, 2000

34

[12] Niedringhaus, W., “A Simulation Tool to Analyze the Future Interaction of Airlines and the U.S. National Airspace System,” Handbook of Airline Strategy: Challenges and Solutions, Regulatory issues, and Public Policy, edited by Butler, G. F. and Keller, M. R.. New York, NY: McGraw-Hill, pp. 619-627, 2001

[13] Niedringhaus, W., “The Jet:Wise Model of National Airspace System Evolution,” Simulation, vol. 80, no. 1, pp. 45-58, 2004

[14] Ryan, A. J., Increasing Route Utilization via User Fees: An Agent Based Model.

[15] Shah, A. P., Pritchett, A. R., Feigh, K. M., Kalaver, S. A., Jadhav A., Corker, K. M., Holl, D. M., Bea, R. C., “Analyzing air traffic management systems using agent-based modeling and simulation,”, Georgia Institute of Technology, Atlanta, GA/San Jose State University, San Jose, CA/ATAC Corporation, Sunnyvale, VA

[16] Sweet, D. N., Manikonda, V., Aronson, J. S., Roth, K., Blake M., “Fast-time simulation system for analysis of advanced air transportation concepts,” AIAA Modeling and Simulation Technologies Conference and Exhibit, Monterey, CA, 2002

[17] Wojcik, L. A., “Models to Understand Airline and Air Traffic Management Authority Decision-Making Intersections in Schedule Disruptions: From Simple Games to Agent-Based Models,” Handbook of Airline Strategy: Challenges and Solutions, Regulatory issues, and Public Policy, edited by Butler, G. F. and Keller, M. R.. New York, NY: McGraw-Hill, pp. 549-575, 2001

[18] Wojcik, L. A., “Airline personalities and air traffic flow management: a simple agent-based model,” AIAA 4th Aviation Technology, Integration and Operations (ATIO) Forum, Chicago, IL, 2004

[19] Wojcik, L. A., “Modeling Distributed Human Decision-Making in Traffic Flow Management Operations,” 3rd USA/Europe Air Traffic Management R&D Seminar, Napoli, 2000

[20] Wojcik, L. A., “Three Principles of Decision-Making Interactions in Traffic Flow Management Operations,” 4th USA/Europe Air Traffic Management R&D Seminar, Santa Fe, 2001

[21] Williams, A., “Airport Demand/Capacity Model,”, The MITRE Corporation, Mclean, VA, 2005

[22] Zelinski, S. J., “Validating The Airspace Concept Evaluation System Using Real World Data,” AIAA Modeling and Simulation Technologies Conference and Exhibit, San Francisco, CA, 2005

[23] Zhang, L. and Levinson D., “An Agent-Based Approach to Travel Demand Modeling: An Exploratory Analysis”

[24] Coldren, G. M., Koppelman, F. S., Kasturirangan, K. Mukherjee, A., “Air Travel Itinerary Share Prediction: Logit Model Development at a Major U.S. Airline,” 82th Annual Transportation Research Board, Washington, D.C., 2003

[25] Adler, T. Falzarano, C. S. and Spitz G., “Modeling Service Trade-Offs in Air Itinerary Choices,” Transportation Research Board: Journal of the Transportation Research Board. No. 1915. Washington. D.C., pp. 20-26, 2005

35

[26] Mashford J. S. and Marksjo B. S., “Airline Base Schedule Optimization by Flight Network Annealing,” Annals of Operations Research 108, pp. 293-313, 2001

[27] Pate. H. R., McDonald B., Gillespie, H. W., “Congestion and Delay Reduction at Chicago O’Hare International Airport,”, 2005

[28] Grosche T., Rothlauf F., “Air Travel Itinerary Market Share Estimation,” Working Paper, Mannheim/Germany, 2007

[29] Pate. H. R., McDonald B., Gillespie, H. W., “Congestion and Delay Reduction at Chicago O’Hare International Airport,”, 2005

[30] Smith, P., “Code Sharing Between Airlines,”, http://www.bootsnall.com/guides/05-06/code-sharing-between-airlines.html

36

3 International Enplanements within the Continental U.S. (to be submitted to Journal of

Air Transport Management)

Abstract






domestic enplanements. The objective of this chapter is to forecast the international passenger

traffic between the CONUS and the rest of the world (international + non-CONUS, i.e., Hawaii,

Alaska and the U.S. territories) between 2009 through 2040.

Nine linear regression models are developed to forecast the enplanements from the United States

to each of nine international regions. The regions include Europe, Asia, Africa, South America,

Mexico, Canada, Caribbean and Central America, Middle East, and Oceania. The international

enplanements from the CONUS to each world region are modeled as a function of GDP and

GDP per capita of both the United States and the specific region. A dummy variable is also used

to account for the effects of September 11, 2001.

The resulting regression models are statistically valid and have parameters that are credible in

terms of signs and magnitude. The total number of international enplanements is forecast to

increase from 74.7 million in 2008 to 184.4 million in 2028. The average annual growth rate is

expected to be 4.7%.

The model described in this chapter has been integrated into the Transportation Systems

Analysis Model (TSAM) developed by the Air Transportation Systems Laboratory at Virginia

Tech. TSAM was originally calibrated to predict domestic origin-destination passenger flows

between 3,091 counties within the CONUS. The modeling effort described in this analysis

addresses a critical gap in the TSAM model.

37

3.1 Introduction


the United States. The share of international air passenger traffic relative to total air passenger

traffic increased in terms of enplanements, seat capacity and aircraft departures between 1990

and 2008. According to Federal Aviation Administration (FAA) statistics, international

enplanements in the United States grew by 21.4 million between 2000 and 2008 with an annual

average growth rate of 4.1% comparing with that of 0.7% for domestic enplanements. The

objective of this chapter is to develop econometric models to forecast the international passenger

traffic between the continental U.S. (CONUS) and the rest of the world (international + non-

CONUS, i.e., Hawaii, Alaska and the U.S. territories) between 2009 through 2040.

The international air travel demand within the CONUS has grown considerately over the period

of 1990-2008. Figure 3-1 shows the evolution of the number of enplanements, seat capacity and

aircraft departures for both domestic (between CONUS) and the international (between CONUS

and international regions plus Hawaii, Alaska and the U.S. territories) air travel. During 1990

and 2008, the number of international enplanements, seat capacity and departures grew by 106%,

69% and 121%, respectively. The growths are much higher than the corresponding growth (51%,

11% and 46%) of domestic air travel.

Because of the higher development rate of the international air travel, the share of international

travel over total travel demand increased in terms of the number of enplanements, seat capacity

and aircraft departures between 1990 and 2008. As is evident in Figure 3-2, the share of

international enplanements increased from 9.2% to 12.1%. The share of international seat

capacity increased from 8.1% to 11.9%. The share of international aircraft departures increased

from 5.1% to 7.5%.

Figure 3-3 shows the share of international enplanements relative to the total enplanements at 15

airports with most international enplanements between 1990 and 2008. In the past 19 years, the

share of international airline passenger enplanements experienced a noticeable growth at the

majority of the top 15 airports. It should be noted that at the Miami International Airport, the

international enplanements share is above 50% in 2008. Historical trends show that a high

proportion of international passenger traffic is concentrated at large international airports within

38

CONUS between 1990 and 2008. These 15 airports (22%) among 69 airports having more than

2,500 international enplanements cover together 84% of total number of international

enplanements in 2008.

Figure 3-1: Evolution of International Enplanements/Seats/Departures within CONUS (Data

Source: 1990 - 2007 T100 International Market Data).

80

100

120

140

160

180

200

220Inde

x of Enp

lane

men

ts

(1990 = 100)

Year

CONUS ‐ (Int'l + US Territories)

CONUS ‐ CONUS

80

90

100

110

120

130

140

150

160

170

180

Inde

x of Seats

(1990 = 100)

Year


CONUS ‐ CONUS

80

100

120

140

160

180

200

220

240

Inde

x of Dep

artures

(1990 = 100)

Year


CONUS ‐ CONUS

39

Figure 3-2: Evolution of Share of International Enplanements/Seats/Departures over

Corresponding Total Values within CONUS (Data Source: 1990 - 2007 T100 International

Market Data).

Figure 3-3: Share of International Enplanements over Total Enplanements at 15 Airports with

Most International Enplanements (Data Source: 1990 - 2007 T100 Domestic/International

Market Data).

4%

5%

6%

7%

8%

9%

10%

11%

12%

13%

Share of 'CONUS ‐(Internationa

l +

U.S. Territories)' ov

er 'CONUS ‐A

ll'

Year

Enplanements

Seats

Departures

0%

10%

20%

30%

40%

50%

60%

70%

JFK LAX MIA ORD EWR ATL SFO IAH IAD SEA DFW PHL DTW BOS FLL

Share of Enp

lane

men

ts to

(Interna

tional + US

Territories) over A

ll En

planem

ents in

2008

15 Gateway Airports with Most Int'l Enplanements in 2008

Year 1990

Year 2008

40

Combined with the analysis of demand for domestic air travel, the forecast of international air

passenger demand can be used for policy making, airport planning, marketing, and investment

decision making. For example, total (domestic plus international) flight operations can be used in

the capacity-delay analysis of an airport. Since the aircraft size of international flights is

generally larger than that of domestic flights, the impact of an international flight on the airport

operation is relatively larger than that of a domestic flight.

Considering the greater impact of international flights on airport operations and the fast growth

of international air travel along with its proportion relative to the total air traffic at airports in

U.S., it is essential to include international operations in any type of airport analysis, such as

capacity-delay analysis, and in future airport expansion plans. Few publicly-available methods

exist to forecast international air travel demand in the U.S.

Both time series methods and econometric models have been used to analyze air travel demand.

The time-series method forecasts the future value of dependent variables applying statistical

analysis only to historical (or time series) data of the variable. Time-series analysis assumes that

one may forecast the value of a variable by studying only the historical pattern of that variable

over time. Time-series analysis is known to be very effective in predicting short-term forecasts

such as monthly, weekly, daily or hourly variations in demand. Though time series techniques

are widely used to model air travel demand, they fail to identify the factors affecting air travel

demand. On the other hand, econometric models assume that the dependent variable to be

forecasted can be explained by the behavior of another or by a set of independent variables. The

purpose of the econometric model is to discover the form of the relationship (mathematical

curve) between all variables through statistical analysis, and to use it to forecast future values of

the dependent variable. The objective of this chapter is to develop econometric models that relate

growth in international air travel demand within the CONUS with the growth of the causative

factors.


3.2.1 Modeling of Air Travel Demand

A variety of models have been developed to forecast air travel demand using econometric

models. All of the models reviewed established an analytical relationship between air travel

41

demand and a set of variables including socioeconomic factors (GDP, population, income, etc.)

and service-related factors (airfare, flight frequency, aircraft size and etc.). They are summarized

in Table 3-1. The previous air travel demand analysis mentioned earlier will be discussed in this

section.

42

Table 3-1: Summary of Previous Passenger Demand Models.

Author and Year

Dependent Variables

Market Segments Independent Variables Demand Functional Form

Data

J.D. Jorge-Calderon 1997

Intra-European, international city pairs with at least 52 flights a year Distance

3 by route distance - < 600 - 601 - 1200 - > 1200

- Population of both origin and destination

- Average distance - Income - Frequency

Semi-log linear

339 cross-section

J. Dargay and M. Hanly 2001

International air travel from/to UK to 20 counties

4 by trip purpose and residents: - UK residents leisure - UK residents business - Non-UK residents leisure - Non-UK residents business

- Real airfare - Disposable income

per capita - Relative price level - Relative exchange

rate

Autoregressive distributed lag

10 years time-series cross-section

InterVISTAS Consulting Inc. 2007

Europe to/from North America

- U.S. domestic - 6 world regions

• Intra-Europe • Intra South Asia and South East

Asia • Trans Atlantic (Europe to/from

North America) • Trans Pacific (South Asia and

South East Asia to/from North America)

• Intra Sub Sahara Africa (Central/Western Africa and Eastern Africa)

• Intra Latin America (Non-Caribbean Central America and South America)

- UK outbound

- Average fare - Geometric mean of

city/country pair GDP

- Geometric mean of city/country pair population

- Route distance - Time period

dummy (quarter/month)

Semi-log linear Time-series cross-section

43

• Leisure • Business

Transports Canada 2003

340 zone pairs 4 by fare class and distance - Domestic economy - Domestic discount - Transborder economy - Transborder discount

Log linear 7 years time-series cross-section

B. Battersby and E. Oczkowski 2001

Revenue passenger kilometer (RPK) per capita of 4 domestic city pairs in Australia

3 by fare class: - Discount economy - Full economy - Business

- Price - Income - Substitute prices - Seasonality dummy

Linear 26 quarters time-series

L. Castelli, R. Pesenti, W. Ukovich (2005)

Daily passenger for 9 routes of one Italian regional carrier

2 by fare class: - Economy - Business

- Population - GDP per-capita - Frequency of

flights - Airfare - Aircraft seat

capacity - Year - Hub Airport - Tourist market - Direct competition - Weekend - May

Semi-log linear

Daily time-series cross-section

S. Y. Abed, A. O. Ba-Fail, S. M. Jasimuddin 2001

International air travel demand (passengers) in Saudi Arabia

N/A - Population - Total Expenditures

Linear 24 years time series

44

Jorge-Calderon (1997) modeled the schedule airline demand for international European routes

considering the geo-economic characteristics of both end-point cities and the airline service

pattern. The geo-economic variables include route distance, population, income, etc. The airline

service related variable included flight frequency, aircraft size and airfare. The model is

formulated as:

ln

ln ln ln

ln ln ln

1 2

1 2

where

Distance between the two end-point cities of route

Summation of the population of the two end-point cities of route

Overall population-weighted average income in both end-point cities of route

Total number of return flights per week on route

Average aircraft size on route

Cheapest unrestricted economy fare available on route

Dummy variable describing the availability of moderately discounted

restricted fares on route

Dummy variable describing the availability of highly discounted restricted fares

on route

Dummy variable describing if route flies over sea water

1 / 2 Dummy variable describing if one/both end-point cities of route are

within 200 km from a major hub

Dummy variable describing if the destination of route is a holiday resort

1 / 2 Dummy variable describing if one/both end-point cities of route are the

hub of a major airline

Coefficient of the th quantitative variable

Coefficient of the th qualitative variable

Constant term

Random error normally distributed with mean = 0 and a constant variance

45

, and were included in the model as endogenous. Two stage

least squares approach was used to estimate the model. The market was segmented by the route

distance ( 600 km, 601 – 1200 and 1200) and market density ( 50,000 km, 50,001 –

100,000 and 100,000). The model was estimated for each of the nine market segments. It was

found that the flight frequency performed better than the average aircraft size to generate traffic

on a shorter route, while the average aircraft size generated more traffic on a longer route. The

highly discounted restricted fare generated traffic on a short route and the demand was inelastic

with respect to the unrestricted economy fare.

Passenger Origin-Destination Model (PODM) is a model Transport Canada used to forecast air

travel demand between defined zones in Canada since 1976 (Transport Canada 2003). The

function form of the PODM model is given by:

ln ln

where

Air passenger trips from zone to zone for market segment

Vector of explanatory variables associated with zone and zone or market

segment

Random error associated with zone and zone for market segment

Vector of parameters to be estimated

PODM-V2 is a new version of the model PODM developed by Tecsult Inc. as a result of

Transport Canada’s request to improve the quality of the forecasts. The model is more general

using a more flexible functional form through Box-Cox transformation as:

where

1, if 0

ln , as 0

1, if 0

ln , as 0

46

The model was calibrated separately for four markets segmented by trip purpose and air ticket

class. The market segments include domestic-economy fare, domestic-discount fare, transborder-

economy fare and transborder-discount fare. Air fare, availability of direct flights, travel time by

car and socio-economic variables including origin zone population, GDP of both origin and

destination zones and origin zone per capita real personal disposable income, and two dummy

variables for year 2000 and 2001 were included in the model as explanatory variables.

Bhadra (2003) related the air travel demand between Metropolitan Statistical Areas (MSAs) in

the U.S. with local area economic and demographic information. It was concluded that local area

information was better explanatory variable to forecast origin and destination travel in

comparison to national area information. Air travel demand was modeled separately for 11

markets segmented by the non-stop distance between origin and destination MSAs:

ln ln ln ln ln ln

ln ln ln

ln

ln

where

One-way average fare between - market

/ Per capita personal income at / MSA

/ Density (per sq mile) at / MSA

/ Multiplicative interactions between population and income

as a measure of degree of economic activities at / MSA

/ Market power (%) of dominant/non-dominant

airlines at the - market

/ Dummy variable of presence of Southwest Airlines as

major/minor airlines at - market

/ Dummy variable for hub status of

origin/destination

Distance between - market

47

Quarter of year

Random error normally distributed with mean = 0 and a constant variance

Dargay (2001) developed four models to estimate UK residents’ leisure/business trips to 20

countries and non-UK residents’ leisure/business trips to the UK individually. UK residents’

leisure/business trips to the 20 countries were modeled as a function of the airfare,

income/foreign trade, the relative price level, the relative exchange rate and previous year trips:

ln ln ln ln ln ln

ln ln ln ln ln ln

where

/ UK residents’ long-run equilibrium demand for leisure/business travel in term

of trips per capita in year

/ UK residents’ leisure/business real air fare between the UK and country in

UK prices in year

UK per capita disposable income in year

Trade between the UK and country in year

Relative price level of country to the UK in year

Relative exchange rate of country to the UK in year

The income of each analyzed country instead of UK income, and local currencies were used to

model the non-UK residents’ leisure/business trips from/to the UK. 10 years of data was used to

calibrate the model and the results showed that income/trade had the greatest impact on air travel

demand.

Rengaraju and Thamizh Arasan (1992) modeled the city-pair air travel demand in India using six

demand variables as shown in Table 3-1, and two supply variables including distance between

the cities and the frequency of service. Additionally, ratio of rail to air travel time was included

as an explanatory variable because of prevailing competition between rail and air in India. A big

city proximity dummy variable was used to consider the reduction in air travel of the small cities

due to their proximity of big cities.

Table 3-2: Socioeconomic Factors, Chosen Function Forms and Assigned Notations of Demand

Variables (Rengaraju and Thamizh Arasan, 1992).

48

The authors used stepwise multiple linear regression analysis to calibrate the model with 40 city-

pairs data. All the estimated coefficients were found to be statistically significant. The estimated

model was given as:

ln 0.595 0.445ln 1.204ln 0.598ln 0.347ln 0.355ln

0.431ln 0.520

where

Ratio of rail to air travel time (line-haul time for rail plus two hours divided by line-

haul time for air plus four hours)

Distance between the two cities in the city pairs

Number of available flights, both ways, per week

Dummy variable of big city proximity ( = 0 for small cities in the proximity of big

cities; = 1 for all the other cities)

Since the model was calibrated with one year of cross-section data, cross validation and

backward prediction were performed before the estimated model was applied to the future years.

Cross validation was performed to estimate the degree of shrinkage of multiple correlation.

Shrinkage was found to be small. Backward prediction was performed to verify the assumption

that the estimated impact of each independent variable on the air travel demand would not

change in the future. The predicted values with a time gap of five years were compared with the

actual values and found to be close to the actual values.

49

3.2.2 Modeling of Air Travel Demand by FAA, Airbus and Boeing

The advantages of econometric models in forecasting long-term trends make them very attractive

and therefore have been adopted in modeling the air travel demand by Federal Aviation

Administration (FAA) and large aircraft manufacturers such as Boeing and Airbus. These

organizations have published their forecasts of international aviation operations. In the

publications, the air travel demand is usually modeled as a linear function of traditional

macroeconomic variables such as GDP and trade [1].

The Terminal Area Forecast (TAF) [2] is a detailed FAA forecast planning database produced

by the Office of Aviation Policy, Planning and Environment (APO). The forecast covers all

airports included in the National Plan of Integrated Airport System (NPIAS). It is updated every

year and the most updated TAF database (February 2006) spans from 2005 to the year 2025. The

TAF predicts air traffic demand at airports using a top-down framework. First, the national-level

passenger enplanements are forecasted using macroeconomic variables, such as income and

population and energy price combined with a time-series model. In a second step, the estimated

national passenger enplanements are allocated to specific airports considering the historical

shares of the airport, master plans and expert opinion [3].

The FAA also provides forecasts of annual international passenger demand between the United

States and the rest of the world over a 12-year forecast period. Every year, the FAA publishes the

“FAA Aerospace Forecast” [1], which contains both historical and forecasts of future

international aviation demand for mainline commercial air carriers, commuter airlines and

general aviation. International aviation demand includes the total number of international

passengers traveling to and from the United States to four world areas: Atlantic, Pacific, Latin

American and Canada. This forecast is an international air travel demand at the national level.

No airport-specific international demand is included in the analysis. Passengers, Revenue

Passenger Kilometers (RPK) and Available Seat Kilometers (ASK) for each area of the world are

predicted. Future load factors and average seats per aircraft of the international flights are also

included in the report. The projection of world economic data such as GDP and exchange rates

of U.S. and various world areas are considered in the analysis. The historical international

passenger data used in the report is collected from United States Immigration and Naturalization

Services (INS) and Transport Canada. The economic data is obtained from Global Insight.

50

Unfortunately, the functional form of the forecasting model used in the “FAA Aerospace

Forecast” is not publicly available. However, it can be inferred from the data tables included in

the report that the international passenger demand prediction is based on an econometric model

using historical passenger statistics and both historical and projected world economic data.

In contrast to FAA, aircraft manufacturers, such as Boeing and Airbus focus on passenger traffic

prediction of RPK and fleet mix rather than on passengers and operations. Every year, Boeing

and Airbus publish individual assessments of the world air travel demand, however neither of

their forecast models is published. As aircraft manufacturers, their emphasis is to predict future

aircraft fleet mixes required to satisfy domestic and international air travel demand. The

“Current Market Outlook” [13] and the “Global Market Forecast” [14] represent the views of

Boeing and Airbus to predict annual air travel demand over a span of 20 years. Boeing updates

their forecast every year, whereas the latest Airbus Global Market Outlook was released in

December 2004. Boeing and Airbus forecasts share a common vision of strong air travel demand

growth over the whole forecast horizon. In the “World Global Market”, Airbus models air

passenger demand based on different sets of economic and air transport variables. The

projections of economic growth and other indices are obtained from the Global Insight

Forecasting Group. World passenger traffic is forecasted to increase 5.3% annually and the

number of flight operations is expected to double during the period 2004-2023. Boeing

considers Gross Domestic Product as the major contributor to the growth in air travel demand.

Other determinants of air travel growth are lower fares, additional world trade, and an increase in

frequencies and more direct service. In the Current Market Outlook (2004), passenger traffic is

expected to grow by 5.2% per year, and World GDP is forecasted to grow by 3.0% annually over

the next 20 years. In terms of the forecast of fleet mix, Airbus and Boeing hold opposite views.

Airbus predicts that:

“Unlike passenger airlines under pressure to improve service levels by increasing

frequencies, freight operators generally have little incentive to increase frequencies beyond

once-daily service and are more likely to respond to growing traffic by increasing aircraft

size and thereby achieving lower unit operating costs”.

51

Boeing on the other hand says that “Airlines will provide passengers point-to-point service on

busy routes” and “Airlines will maintain or reduce airplane size to provide frequent, nonstop

service”.

3.3 Methodology

The international enplanements from the CONUS to the rest of the world are forecasted

separately for 12 regions including nine international regions and three non-CONUS regions. As

listed in Table 3-3, each region is grouped by the world area code. The international regions

include Europe, Asia, Africa, South America, Mexico, Canada, Caribbean and Central America,

Middle East, and Oceania. The non-CONUS regions include Alaska, Hawaii and the U.S.

territories.

Table 3-3: Definition of 12 Regions by World Area Code (WAC).

Region Min. WAC

Max. WAC Remark

International

Africa 501 599 - Asia 701 799 Exclude 770 (Russia) Canada 900 961 - Caribbean & Central America 101 299 Exclude 148 (Mexico) Europe 401 499 Include 770 (Russia) Mexico 148 148 - Middle East 601 699 - Oceania 801 899 - South America 301 399 -

Non-CONUS Alaska 1 1 - Hawaii 2 2 - U.S. Territories 3 5 -

52

Figure 3-4: Enplanements within the CONUS to each of 12 Regions in the analysis (Data Source:

2008 T100 International Market Data).

Figure 3-4 shows the enplanements within the CONUS to each of the 12 regions in year 2008.

The regions are in decreasing order of the number of enplanements from the bottom to the top.

Over 25 millions international enplanements within the CONUS were to Europe. It led the share

(30%) of total international enplanements to 12 regions, followed by the Caribbean & Central

America (13%), Canada (12%), Mexico (11%) and Asia (10%).

Nine linear regression models are developed to forecast the enplanements from the CONUS to

each of the nine international regions. Nine international regions cover together 86% of the total

international enplanements within the CONUS. The number of enplanements from the CONUS

to each of three non-CONUS regions is forecasted by applying their historical average annual

growth rate during the period 1990-2008.

‐ 5 10 15 20 25 30

EuropeCaribbean & Central America

CanadaMexico

AsiaHawaii

South AmericaU.S. Territories

AlaskaMiddle East

OceaniaAfrica

Enplanements within the CONUS in Year 2008 (Millions)

12 Regions in

Ana

lysis

53

Figure 3-5: Evolution of Enplanements within the CONUS to each of the Nine International

Regions (Data Source: 1990-2008 T100 International Market Data).

Figure 3-5 shows the evolution of the enplanements within the CONUS to each of the nine

international regions during the period of 1990-2008. The regions are in decreasing order of the

growth from the top to the bottom. The growth of all regions has been steady, although decreases

occurred in 1991 and during the period 2001-2003. The first reduction in international air

transportation demand is linked with the first Gulf war. The second reduction in demand reflects

the impacts of Sep. 11, 2001. Because of their underdeveloped markets, Africa and Middle East

experienced the most significant growths (433% and 342%, respectively) compared with the

other seven regions. Conversely, the growth of the more mature markets in Europe and Canada is

relatively lower.

The methodology framework of the regression analysis is shown in Figure 3-6.

80100120140160180200220240260280300320340360380400420440460480500520540560

Inde

x of Enp

lane

men

ts within the CO

NUS

Year

AfricaMiddle EastSouth AmericaOceaniaCaribbean & Central AmericaAsiaMexicoEuropeCanada

54

Figure 3-6: Methodology Framework Used to Forecast International Air Travel Demand.

3.3.1 Scatterplot and Correlation Matrix

Before running the regression analysis, a series of scatter plots of each independent variable

against the dependent variable are examined to identify their relationship [16]. The graphical

analysis is helpful in identifying the relationship between the two variables. Additionally, the

graphical analysis provides certain clues with regard to the need for data transformation. Such

information is not easily found when looking at the data in tabular form. If the regression

analysis is performed without graphical analysis, significant relationships between the

transformed data may be missed.

The second step is to obtain a correlation matrix between explanatory variables. Running

correlations among the independent variables is helpful in preventing multicollinearity problems

(multicollinearity will be addressed in the model evaluation section later in this chapter). The

bivariate correlations procedure in SPSS is used in this analysis to compute Peason’s correlation

coefficient between all the independent variables. The correlation coefficient measures how

Develop Regression Model- Scatter Plot- Multicollinearity Test- Assume Functional Form

Evaluate Models- Regression Coefficients- Residual Analysis- R Square

Identify Candidate Variables & Data Collection

Forecast Int. Air Pax. Trips- Forecast Independent Variable

Evaluate Forecast Results- Consistent with Historical Trend- Compare with FAA Forecast

55

variables are linearly related. It ranges in value from -1 (a perfect negative relationship) and +1

(a perfect positive relationship). A value of 0 indicates no linear relationship.

3.3.2 Assumed Functional Form of the Model

The international enplanements from the U.S. to each international region are modeled as a

function of GDP or GDP per capita of both the U.S. and the specific region. Based on the

assumption that the events of the Sep. 11, 2001 terrorist attack caused a permanent reduction in

international air passenger trip, a dummy variable is defined as 1.0 throughout the forecast period

since the year 2001, otherwise, it is zero. The estimated model is displayed in the following

equations.

α Equation 3-1

α Equation 3-2

α Equation 3-3

α Equation 3-4

where:

: Total passengers on commercial flights from the U.S. to

international region i

: Sum of international region i’s GDP and U.S. GDP in year 2000

millions of dollars

: Product of international region i’s GDP and U.S. GDP in year

2000 millions of dollars

: Sum of international region i’s Per Capita GDP and U.S. Per

Capita GDP in year 2000 millions of dollars

: Product of international region i’s Per Capita GDP and U.S. Per

Capita GDP in year 2000 millions of dollars

_ . : Dummy variable explaining 9.11 effect on traffic (0 for 1990 to

2000, 0.25 for 2001 and 1 for 2002-2007)

_ : Dummy variable explaining 9.11 effect on traffic (0 for 1990 to

2000, and 1 for 2001-2007)

α : Estimated intercept

56

, : Estimated coefficients for international region i model

: Random error associated with model for international region i

The coefficients of the explanatory variables are calibrated by linear regression analysis using

the SPSS software program [16].

3.3.3 Model Evaluation

After the model is calibrated, it is of critical importance to perform statistical tests to support the

assumptions made during the regression analysis. All regression coefficients need to be

examined statistically for significance. Additionally, the forecasting accuracy of the model

should be evaluated. All related tests are introduced in the following sections.

3.3.3.1 Tests of Assumptions Underlying Regression Analysis

As a first step in analyzing the regression output, residual analysis is conducted to test the

underlying assumptions in the regression models. The classic ordinary least square procedure for

estimating the regression coefficients for each independent variable assumes the following [6]:

- The forecast errors are normally distributed with a mean of zero.

- The forecast errors are statistically independent of each other (no autocorrelation).

- The variance of the forecast error is constant across all observations and values of

independent variables (homoscedasticity).

The regression analysis assumes that the forecast errors or residuals are normally distributed with

a mean of zero. The forecast residual is defined as the difference between (the actual value

of dependent variable) and (the value predicted by the estimated regression equation). This

assumption implies that, on large samples, the histogram of forecast errors should follow the

pattern of a normal distribution centered at zero. If the histogram is observed as severely skewed,

the violation of the assumption should be investigated. One typical method to verify whether the

forecast errors are normally distributed is the use of normal probability plots. A normal

probability plot is a scatter diagram of the forecast errors over the standard residuals, which are

calculated by:

e

ii S

eZ = Equation 3-5

57

iii YYe∧

−= Equation 3-6

where:

ie = residual of the ith observation

eS = standard deviation of the estimated residuals

If the forecast errors follow a normal distribution, a straight line should be observed in the

probability plot of residuals. Another method for normality test is to compute the Jarque-Bera

statistic. Further details regarding this test can be found in [17]. The histogram combined with

normal probability plot of forecast errors is adopted in this analysis using the linear regression

procedure available in SPSS.

Ordinary least squares analysis also assumes that the individual value of the dependent variable

is not affected by each other, i.e. no autocorrelation. The violation of this assumption frequently

occurs in regression models using time-series data. Autocorrelation of the residuals causes the

estimate of the residuals variance to be small; therefore an unreliable evaluation may result [6].

One widely used method to check the presence of the autocorrelation is the Durbin-Watson test.

The Durbin-Watson (D-W) d-statistic is calculated to determine whether the correlation between

residuals is statistically equal to zero. The range of the D-W d-statistic is always within 0 and 4.

Values close to 2 indicate that autocorrelation does not exist. Values close to 0 and 4 indicates

positive and negative autocorrelation, respectively. More information regarding decision of rules

for the D-W test can be found in [18]. Autocorrelation in the residuals can also be examined by

the plotting of residuals versus time (called time plot of residuals). The time plot of residuals

should not display any discernible trend if autocorrelation is not present. Both the D-W test and

the time plot of residuals are available in SPSS. The D-W test is used in this analysis.

The last assumption in supporting regression analysis to yield valid least-square estimates is the

constant variance of residuals, i.e., homoscedasticity. Residuals with non-constant variance are

said to be heteroscedastic, which distorts the measure of unexplained variation, thus misleading

the standard goodness-of-fit statistic 2R and t-tests (both of them will be introduced later in this

chapter) [6]. Heteroscedastity can be tested by plotting residuals either over time or against the

58

estimated values of the dependent variable. The latter method is employed in this analysis. If the

variances show no discernible pattern, then the homoscedasticity assumption is verified.

3.3.3.2 Statistical Tests of the Regression Coefficients

Once the regression model for international air passenger forecasts has been calibrated and tested

for the underlying assumptions of the linear regression models, a graph of the observed data vs.

the forecasted value is generated to determine how well the model works. It is also necessary to

perform statistical tests to examine the forecasting accuracy of the model.

Another important task to evaluate the regression model is to examine the statistical validity of

the regression coefficients. Appropriate t-tests for each regression coefficient should be

performed to determine the level of statistical significance. Each coefficient should be analyzed

to check whether the estimated sign of the coefficient is consistent with the expectations of the

sign. It is also appropriate to evaluate the magnitude of the coefficient to make sure that it is

reasonable within the context of the computations.

A T-test is used to examine whether each independent variable contributes significantly (i.e.

statistically significantly different from zero) to the model in a statistical way. Typically, the t-

value greater than 2 is considered statistically significant from zero. Therefore statistically

meaningful relationships between the dependent variable and the corresponding independent

variables can be quantified by the examination of the appropriate t-value.

The estimated sign of the coefficient should be consistent with the logical expectations of the

sign. For example, according to economic theory, the coefficient for GDP in air travel demand

model should be positively related. Similarly, the magnitude of the coefficient should be

reasonable within the context of the model. Any explanatory variable with a “wrong sign”, or a

“wrong magnitude” or any explanatory variable considered to be “statistically insignificant” can

be dropped from the regression equation. As previously explained, multicollinearity and

violation of the constant variance assumption may cause the “wrong sign” of the regression

coefficient. Additionally, the “wrong magnitude” might be caused by an error in the data

tabulation. It is not surprising that in many cases, a regression equation with an “insignificant”

explanatory variable frequently provides a better prediction after the insignificant explanatory

59

variable is omitted. Thus, these issues should be considered when select variables in the final

regression equation.

3.3.3.3 Multicollinearity

Multicollinearity is also important to test during the model evaluation because its existence may

inflate the variances of the parameter estimates. The multicollinearity occurs when two or more

explanatory variables are highly correlated with each other, leading to unreliable parameter

estimates. Consequently, the multicollinearity makes the interpretation of the estimated

parameters extremely difficult. Usually the explanatory variables with t-statistics below the

critical value given the sample size and level of confidence should be excluded from the final

model. The presence of multicollinearity creates lower t-statistic values because the standard

errors of the regression coefficients are largely overestimated. Without a multicollinearity

evaluation, it is no longer possible to determine which of the independent variables are relevant

in the regression equation.

The identification of the multicollinearity problem focuses on determining its seriousness more

than presence [19]. Independent variables are almost always correlated to some degree, so the

influence of multicollinearity on the regression results is a matter of degree. Multicollinearity can

usually be identified by the following three approaches: The first approach is to check the logical

correlation coefficients between the dependent and independent variables. Significant but

‘wrong’ sign of regression coefficient between the dependent and independent variables show

severe multicollinearity. A second method is to estimate the correlation matrix for independent

variables. High correlation coefficients (0.8 or higher) indicate serious multicolinearity problem.

The third approach is to examine the different values of 2R with leaving out each of the

explanatory variables. The small difference change in 2R indicates a significant multicollinearity

or values of the Variance Inflation Factor (VIF, SPSS has it as a multicollinearity diagnostic

statistic) exceeding 10 can indicate multicollinearity [20]. In this analysis, a combination of these

three methods is used to test multicollinearity. The effect of multicollinearity can be reduced by

increasing the sample size and eliminating redundant independent variables. For this reason, the

U.S. GDP is excluded from the regression models developed. The U.S. GDP and World Region

GDP are found correlated with the dependent variable during the model development.

60

3.3.3.4 Testing the Estimated Model for Overall Significance

A standard measure of goodness-of-fit used to evaluate regression model is the statistic along

with the coefficient of determination or . The F-statistic is used to test whether there is a

statistically significant relationship between the dependent and the independent variables (all the

regression coefficients excluding the intercept are zero). The statistic is the ratio of the mean

square regression [SSR/ (k-1)] to the mean square error [SSE/ (n-k)]:

∑

∑

=

=

−−

−−= n

iii

n

iii

kYY

knYYF

1

2

1

2

)1/()ˆ(

)/()ˆ( Equation 3-7

where:

k = the number of estimated coefficients plus the intercept estimated in the regression (for

bivariate regression, it is 2)

n = the sample size

The coefficient of determination can be used to tell how well the model fits the observed data.

Whereas the F statistic is a useful test of the estimated model’s ability to explain any variation in

the dependent variable, it does not provide clues about the strength of the explanatory power.

The value of 2R measures the percentage of the variation in the dependent variable that is

explained jointly by the independent variables . For example, is 0.80, which can be

interpreted as 80 percent of the variation in can be explained by the estimated model. Although

a high does not necessarily mean an “appropriate” model, an “appropriate” model is expected

to have a reasonably “high” [6]. The better the fit of the regression model, the closer is to

1.

The statistic is calculated by comparing the explained variation of the model to the total

variation. It is estimated using the following equation:

∑

∑

=

=

−

−−= n

iii

n

iii

YY

YYR

1

2

1

2

2

)(

)ˆ(1 Equation 3-8

where:

61

iY = observed value

iY = sample mean

iY = predicated value

n = sample size

Instead of standard statistic, the is employed in the study to measure the goodness-of-fit

of the regression model. is always recommended when comparing two alternative models

containing an unequal number of independent variables or two alternative models using different

functional forms but the same sets of independent variables [21]. The value of always

increases as the number of independent variables is increased in a regression equation. In other

words, the goodness-of-fit of the model appears to improve by adding additional independent

variables. Using , the benefits obtained by adding additional independent variables are

balanced against the cost of losing additional degrees of freedom. An adjusted value can be

calculated as follows:

1pn1nR11R

2

adj2

−−−−

−=))((

Equation 3-9

where: 2R = Coefficient of determination

n = the sample size

p = the number of independent variables included in the air travel demand model

3.3.3.5 Evaluation of Model Forecasts over Historical Periods

The root mean square error (RMSE) is a measure of standard deviation of the forecasting error

and is defined as the difference between the observed data and the forecasted value generated by

the model. The RMSE statistic is estimated as follows:

n

eRMSE

n

1t

2t∑

== Equation 3-10

where:

te = the forecast error in time period t

n = the sample size

62

Two types of RMSEs, within-sample and out-of-sample, are usually applied to determine the

forecasting accuracy of a model [6]. “Within-sample” RMSE is calculated for alternative models

that use the entire available historical data. Whereas, “out-of-sample” RMSE can be estimated for

the models whose calibrations are conducted by portioning the whole observed historical data

into two subsets. The first subset is used to calibrate the model and generate the forecast. The

forecast is then used to compare not only with the actual data from the first subsets, but also with

the observed data from the second subset that is not employed in the model calibration. Because

of data limitations, the model presented in this analysis is calibrated with the complete historical

data, thus only “within-sample” RMSE is calculated.

RMSE can be used to select the best forecasting model by simply choosing the model with the

smallest RMSE. However, it is important to notice that comparisons of RMSE for alternative

forecasting models using different transformations of the data are not permissible [22].

3.3.3.6 International Passenger Enplanements Forecast

The forecasts of international passenger enplanements from the U.S. to each World Region are

obtained using individual regression models. During this stage, the collected GDP projections of

each World Region combined with the defined dummy variable for the Sep. 11, 2001 terrorist

attack are input into the resulting model separately.

It is assumed that the historical relationship between the passenger enplanements from the U.S.

to each World Region and each World Region’s GDP may not change over the forecasting

horizon. An econometric model that fits well with the historical data to which it is calibrated

does not guarantee its future forecasting accuracy. New factors that contribute to international air

passenger demand may appear in the future. Based on this uncertainty, it is appropriate to

assume that the World Region GDP will remain as the main variable to explain international air

passenger demand in the future.

3.4 Results

3.4.1 Model Development and Evaluation Results

range from 0.6 to 0.97. The statistical validity and forecast accuracy of all of the regression

models is found to be acceptable throughout the tests discussed in the previous section.

63

Table 3-4 presents the estimated regression equations used to forecast the enplanements within

the CONUS to each international region. The R-squared values for the nine estimated models

range from 0.6 to 0.97. The statistical validity and forecast accuracy of all of the regression

models is found to be acceptable throughout the tests discussed in the previous section.

64

Table 3-4: Regression Equations to Forecast Air Passenger Demand from the U.S. to each of the Nine International Regions.

World Region Selected Regression Equation [t-statistic*] Adjusted

R-Square

Africa 148,769 0.053748GDPUS A 164,329D _ 0.95

Asia 5126271 664.72543 1,606,269 _ 0.96

Canada 2,402,279 0.008404 1,207,242 _ 0.97

Caribbean &

Central America

PUS C & 2,914,336 0.079425PerCapitaGDPUS C &

0.98

Europe 26,277,146 1988.597996 5,735,867 _ . 0.97

Mexico PUS M 781,670 0.781328GDPUS M 0.95

Middle East 134,715 0.089775 290,548 _ 0.93

Oceania 64,441 0.133412 240,852 _ . 0.95

South America 3,838,077 659.143891 _ 1,422,489 _ . 0.92

*: The value within the square bracket is t-statistic value for the regression coefficient.

65

The underlying assumptions of the regression analysis are found to be valid using the

residual analysis for all of the models developed. The tests for these validations are: the

histogram of residuals and normal probability plot of the standardized residuals are

combined to test the assumption of normal distribution with a mean of zero; The plots of

standardized residuals versus the standardized predicted values of each model are used to

verify the homoscedasticity assumption; The critical Durbin-Watson statistic values are

checked to support the nonautocorrelation assumption. Furthermore, the VIF value less

than 10 is helpful in excluding any multicollinearity problems. All regression coefficients

for each explanatory variable are also tested as statistically significant. The signs of each

explanatory variable match the underlying economic theory and expected magnitudes.

Finally, acceptable values and statistically significant F static values prove that all

estimated models are reliable to forecast the passenger enplanements from the CONUS to

each international region.

Since the individual regression model is developed to forecast the passenger

enplanements from the CONUS to each international region, the evaluations are

conducted separately for each model. The regression model and the model evaluation for

Europe are presented here as an example.

The residual analyses for the U.S. to Europe model evaluation are also shown in Figure

3-7, Figure 3-8 and Figure 3-9. Figure 3-7 presents the normal probability plot of

standardized residuals generated by the estimated model. The plot displays the normal

distribution of regression residual because it follows the 45 degree line very well. A

frequency histogram of the same residuals is plotted in Figure 3-8. The figure further

indicates that no skewness occurs about the mean, which satisfies the assumption that the

mean of the regression residual is zero.

To examine the assumption of homoscedasticity, a plot of standardized residuals against

the standardized predicted values is graphically shown in Figure 3-9. No apparent trend is

observed in this graph and all residuals fall within a horizontal band located two standard

deviations from the zero mean.

66

Observed Cumulative Probabilities

Figure 3-7: Normal Probability Plot of Regression Standardized Residual

(Normality Test - Model to Forecast Air Passengers from the U.S. to Europe).

Figure 3-8: Histogram of Regression Standardized Residual

(Zero Mean Test - Model to Forecast Air Passenger from the U.S. to Europe).

Regression Standardized Residuals

Expe

cted

Cum

ulat

ive

Prob

abili

ties

67

Stan

dard

ized

Reg

ress

ion

Res

idua

ls

Observed Cumulative Probabilities

Figure 3-9: Regression Standardized Residual V.S. Standardized Predicted Value

(Homoscedasticity Test - Model to Forecast Air Passenger from the U.S. to Europe).

The lower and upper critical values of the Durbin-Watson statistic for nineteen

observations and two explanatory variables at 5% significance level are 0.835 and 1.264,

respectively. The Durbin-Watson statistic value of 2.29 (1.264 < 2.29 < 2.736 = 4 -

1.264) in the model summary in Table 11 provides the evidence to reject the

autocorrelation associated with the residual test. Finally, a VIF value of 2.96 in the

Coefficients of Table 11 does not imply multicollinearity problem.

To further validate the estimated models, the historical number of enplanements within

the CONUS to each of the nine international regions is compared with the corresponding

estimated numbers using the models. Figure 4-11 shows this comparison for Europe. As

shown, the estimated values follow the historical trend most of time, however

discrepancies do exist between the historical value and the forecast in year 1991 and

2001. The discrepancy in year 1991 was caused by the first gulf war, while the

discrepancy in year 2001 was caused by the 911 event. Comparisons for all other eight

international regions are included in Appendix B.2.

68

Figure 3-10: Comparison between Historical and Forecast Enplanements within the

CONUS to Europe during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International

Market Data).

3.4.2 Forecast Results of Passenger Enplanements from the U.S. to each

International Region

Table 3-5: Adjustment Factors for Estimated International Air Travel Demand Models.

International Region Model Adjustment Factor

1 Africa 33,2082 Asia ‐649,3893 Canada 203,2894 Caribbean & Central America 225,3785 Europe ‐942,5256 Mexico 182,2247 Middle East 214,1128 Oceania ‐95,7629 South America ‐248,517

All of the nine estimated models are applied using the forecast GDP (Appendix B.1: -

Table B-4, B-5) or Per Capita GDP (Appendix B.1: Table B-6, B-7) of the U.S. and each

of the nine international regions from USDA International Maroeconomic Data Set for

the period 2009 – 2040. The resulting forecast of the number of enplanements to each

region is adjusted by their corresponding adjustment factors. The adjustment factor for

each of nine models is the difference between the base year (2008) historical passenger

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20081

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

5-Europe (2008 Market Share: 29.8228%) Passengers from the CONUS 1990 - 2008

Forecast - Linear PerCapitaGDPUS+Europe (R2 = 0.981)

Historical

69

traffic from the U.S. to the specific region and the base year forecast. The adjustment

factors for all of the nine international passenger demand models are listed in Table 3-5.

Figure 3-11: Comparison Between Forecast Growth and Historical Growth.

Figure 3-11 shows the forecasted average growth rate for each region during the period

2009-2040. Aside from Alaska, Hawaii and U.S. Territories that apply historical growth

rate in the future, the forecast growth for the nine international regions are consistent with

their historical growth. For example, the highest growth rates are predicted for Africa at

8.4% per year, and 8.0% for Middle East. These two regions also experienced the highest

historical growth. Their significant growth can be explained by their relatively

underdeveloped air travel demand markets today, comparatively speaking.

As shown in Figure 3-12, the number of enplanements within the CONUS to Europe is

projected to increase at an average annual rate of 3.8% from 25.8 million in year 2008 to

84.1 million in 2040. The number of enplanements is forecast to decrease in 2009 due to

the weak world economy. The forecasted trend is consistent with the historical trend. The

forecast growth rate is slightly lower than the historical growth of 4.1% during the period

1990-2008. The figures showing the forecast trend of each of nine international regions

are included in Appendix B.2 (Figures B-1 to Figure B-9).

0% 2% 4% 6% 8% 10% 12%

EuropeCaribbean & Central America

CanadaMexico

AsiaHawaii

South AmericaU.S. Territories

AlaskaMiddle East

OceaniaAfrica

Growth/Year of Enplanenents within the CONUS

12 Regions in

Ana

lysis

Forecast during 2009‐2040

Historical during 1990‐2008

70

Figure 3-12: Historical (1990 – 2008) and Forecast (2008 – 2040) Enplanements within

the CONUS to Europe (Historical Source: 1990 - 2008 T100 International Market Data).

Figure 3-13: Historical (1990 – 2008) and Forecast (2009 – 2040) Enplanements within

the CONUS to All 12 Regions (Historical Source: 1990 - 2008 T100 International Market

Data).

As shown in Figure 3-13, the total number of international enplanements within the

CONUS to all 12 regions is forecasted to increase from 74.7 million in 2008 to 392.7

million in 2040. The average annual growth rate is expected to reach 4.8%, which is

slightly higher than the historical growth rate of 4.2% during the period 1990-2008.

1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20401

2

3

4

5

6

7

8x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US


Forecast - Linear PerCapitaGDPUS+Europe (R2 = 0.981; Avg. Growth Factor = 3.5026%)

Historical (Avg. Growth Factor = 4.1255%)

1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

0.5

1

1.5

2

2.5

3

3.5

4x 10

8

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

Passengers from the CONUS to Internatioanl + HI, AK & U.S. Territories 1990 - 2040

Forecast (Avg. Growth Factor = 4.5419%)Historical (Avg. Growth Factor = 4.1854%)

71

3.4.3 Evaluation of the Forecast Results

The forecasted trend of international enplanements within the CONUS to each of the

international regions is consistent with the corresponding historical trend. This can be

examined in Appendix B.3 (Figures B-10 to B-18). The forecast of enplanements within

the CONUS to each of the international regions is also compared with FAA aviation

forecasts. It is helpful to compare the forecast results with other models using the same

metric. As stated in Chapter 2, the FAA, Boeing and Airbus produce their own forecasts

of international air travel demand. However, the forecast by Boeing and Airbus focuses

more on the aircraft fleet mix rather than on passenger enplanements for future years. On

the other hand, FAA publishes forecasts of international enplanements for four different

world areas: Canada, Atlantic, Asia Pacific, and Latin America. The FAA forecast of

enplanements to each world area and the passenger enplanements predicted by the

models developed are compared. The average growth rates for Canada and Latin America

areas forecasted by our modes are close to FAA forecasts. Disparities are found between

the predictions for the other two world areas, which may be explained by the assumptions

made by the FAA on values of GDP used.

It should be noted that GDP data used in FAA forecast is different from the GDP forecast

employed in this analysis. The FAA uses INS Form I-92 International Air Travel

Statistics published by the U.S. Department of Commerce as the data source to collect the

passenger enplanement flows from the CONUS to the rest of the world. Our study uses

T-100 international Market data collected by the U.S. Department of Transportation as

the main data source of the dependent variables. The disparities observed between the

FAA predictions and our models can be explained by the different sources of data sets

used.

3.4.4 Model Application


Analysis Model (TSAM) [24] developed by the Air Transportation Systems Laboratory

at Virginia Tech. TSAM was originally calibrated to predict domestic origin-destination

passenger flows between 3,091 counties within the CONUS. It also predicts origin-

destination passenger flows between airports. The framework of TSAM is depicted in

Figure 31.

72

Figure 3-14: Evolution Framework of Transportation System Analysis Model (TSAM).

The international air travel demand within the CONUS has grown considerately over the

period of 1990-2008. During the same time period, the share of total international air

passenger traffic relative to total air passenger traffic increased in terms of enplanements,

seat capacity and aircraft departures. Additionally, the share of international airline

passenger enplanements experienced a noticeable growth at majority of the top 15

airports. The modeling effort described in this analysis addresses a critical gap in the

TSAM model.

3.5 Conclusion

Nine linear regression models are developed to forecast the enplanements within the

CONUS to each of the nine international regions. The regions include Europe, Asia,

Africa, South America, Mexico, Canada, Caribbean and Central America, Middle East,

and Oceania. The international enplanements within the CONUS to each international

region are modeled as a function of GDP and GDP per capita of both the United States

73

and the world regions. A dummy variable is also used to account for the effects of

September 11, 2001.

The resulting regression models are statistically valid and have parameters that are

credible in terms of signs and magnitude. The total number of international enplanements

is forecasted to increase from 74.7 million in 2008 to 184.4 million in 2028. The average

annual growth rate is expected to be 4.7%.


Analysis Model (TSAM) [24] developed by the Air Transportation Systems Laboratory

at Virginia Tech. TSAM was originally calibrated to predict domestic origin-destination

passenger flows between 3,091 counties within the CONUS. The modeling effort

described in this analysis addresses a critical gap in the TSAM model.

To enhance the models presented in this analysis, besides GDP or GDP per Capita, other

variables should be investigated that might increase the value of the regression

models for several regions such as Asia, Oceania and Africa. Globalization, international

trade, declining fares and network development can all be used to explain additional air

travel demand in the future.

74

4 The Impact of the EU-US Open Skies Agreement on Commercial Airline

Passenger Traffic over the North Atlantic (to be submitted to Journal of Air

Transport Management)

Abstract

The European Union – United States Open Skies Agreement became effective March 30,

2008. The objective of this chapter is to develop mathematical models which permit

forecasting the effect of EU-US Open Skies Agreement on commercial airline passenger

traffic over the North Atlantic Ocean. Specifically, models are developed to predict the

passenger traffic amongst the most highly travelled United States to European country

pairs, and to predict the traffic amongst the highly travelled passenger United States

airport to European airport pairs. From these predictions, potential new nonstop flight

airport pairs are offered. Lastly, suggestions are made as to European airports where

selected passengers will be rerouted to address the situation when London, Heathrow

Airport reaches full capacity and can accept no additional passenger traffic.

Nine econometric models were developed to forecast passenger traffic between the

United States and nine selected European countries between 2008 through 2020. The

transatlantic passenger traffic was modeled as a function of the product of the United

States Gross Domestic Product (GDP) and the European country’s GDP with a dummy

variable to account for the effects of September 11, 2001 events. A semi-logarithmic

linear relationship was recommended from modeling purposes.

Sixty-eight new nonstop flights between the United States airports and the European

airports are predicted by the model in 2020 using the airport pair passenger demand

forecast. These new nonstop flights are forecast for two types of airport pairs: (1) pairs

which in 2007 could only be travelled with connecting flights (2) and pairs which in 2007

could be travelled in one flight which includes intermediate stops.

As demand for air travel continues to grow in the future, passenger traffic at certain

airports could place further strain in passenger capacity. Part of the traffic at these

airports will, by necessity, be diverted to other airports or perhaps even choose other

modes of transportation. Diversion of passengers traffic from the United States to

75

London, Heathrow is demonstrated as an example for rerouting the excess air travel

passengers from one airport to other airports when the airport operational capacity is

exceeded assuming all rerouted travelers continue to use the commercial air travel mode.

76

4.1 Introduction

The European Union – United States Open Skies Agreement, which became effective

March 30, 2008, was established to accomplish several goals including:

• “ … to promote an international aviation system based on competition among airlines

in the marketplace with minimum government interference and regulation.”

• “ … to facilitate the expansion of international air transport opportunities, including

through the development of air transportation networks to meet the needs of

passengers and shippers for convenient air transportation services.”

• “ … to make it possible for airlines to offer the travelling and shipping public

competitive prices and services in open markets.”

The objective of this chapter is to develop forecasting models to predict the effect of the

new Open Skies agreement on the behavior of passengers flying across the North Atlantic

Ocean and airline’s behavior such as opening new nonstop flights, increase frequency,

decrease airfare, and adoption the new available aircraft type based on the new aircraft

performance.

Data shows that more than 51 million (25.5 million from the United States to Europe +

25.6 million from Europe to the United States) passengers travelled between Europe and

the United States in 2007. These passengers took direct flights (either non-stop flights

involving no intermediate stops or flights including a stopover at an intermediate airport

without change in flight number) between 31 gateway airports in the United States and 35

European gateway airports. The analysis performed uses a representative number of

European Countries and gateway airports and a representative number of gateway

airports in the United States to predict passenger and airline’s behavior.

4.2 Domain of Analysis

4.2.1 Time Frame of Forecast Model

The time frame of applicability of the model will be from 2008 through 2020.

Socioeconomic data from the U.S. Department of Agriculture is not projected beyond

2020. Therefore, the forecasts are not made beyond 2020.

77

4.2.2 Geographical Domain of Analysis

Along with the United States, nine of the 44 European countries are included in the

domain of analysis for the forecast model. As discussed hereafter, these European

countries contribute the great majority of passengers to North Atlantic air traffic, have

and are projected to have the great majority of the gross domestic product for Europe and

have no missing or abnormal in their historical passenger traffic data.

It is observed from T100 International Market data that passengers crossed the Atlantic

by the direct flights (nonstop flights or flights that have stopover(s) without change in

flight number) between the United States and 24 European countries in 2007. Nine

European countries included in the analysis domain are highlighted in Figure 4-1.

Figure 4-1: 2007 Passenger Traffic from the United States to European Countries

(Source: 2007 T100 International Market Data).

0 2 4 6 8 10

U.K.Germany

FranceNetherlands

ItalyIrelandSpain

SwitzerlandBelgiumDenmarkPolandIcelandRussia

SwedenGreeceAustriaPortugal

CzechUkraineFinlandHungaryNorwayRomania

Latvia

Passengers from U.S. to Europe (Millions)

Included nine European countries

78

Figure 4-2 shows 91% of the total passenger traffic from the U.S. to Europe is included in

the analysis domain when include the European countries with the top nine passenger

traffic in 2007.

Figure 4-2: 2007 Cumulative Percentage of Total Passenger Traffic from the U.S. to

Europe (Source: 2007 T100 International Market Data).

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Cumulative Pe

rcen

tage

of Total Passenger

Traffic

from

U.S. to Eu

rope

European countries with the top nine passenger traffic in 2007 captured 91% of total passenger traffic from U.S. to Europe

79

Figure 4-3: Map of Europe and Nine Selected European Countries in the analysis.

Assumption: It is assumed that the country-to-country passenger traffic between the

United States and the selected nine European countries in the analysis domain is

symmetric. Based on this assumption, only one direction of passenger traffic from the

United States to the selected nine European countries is forecast. The passenger traffic for

the other direction is assumed to be same.

The total passenger traffic, the country-to-country passenger traffic and the airport-to-

airport passenger traffic between the United States and selected nine European countries

are all symmetric. In 2007, total 23.3 million air passengers were transported from the

United States to selected European countries. Comparably, 23.4 million air passengers

were transported from selected nine European countries to the United States. The

symmetry for country-to-country passenger traffic and airport-to-airport passenger traffic

between the United States and nine European countries is seen from Figure 4-4 and

Figure 4-5, respectively. Figure 4-5 only shows the transatlantic airport pairs with 35

0 200 400 600100Miles

:Legend

Selected Nine European CountriesEuropeOther Continents

Atlantic Ocean

Black Sea

Mediterranean Sea

Caspian Sea

Balti

c Sea

North Sea

Arctic Ocean

Arctic Circle

Spain

FranceItaly

Germany

United KingdomIreland

Italy

Italy

Belgium

Switzerland

Netherlands

United Kingdom

-20°-30°-40°

-10°

-10°

0°

0°

10°

10°

20°

20°

30°

30° 40°

40°

50°

50°

60°

60°

70° 80°

40°

40°

50°

50°

60°

60°

70°

70°

80°

80°

80

most passenger traffic, the passenger traffic for more transatlantic airport pairs is

included in Appendix C.1.

Figure 4-4: Country-to-Country Passenger Traffic from Europe to the United States vs.

Traffic from the United States to Europe (Source: 2007 T100 International Market Data).

Figure 4-5: Airport-to-Airport Passenger Traffic from Europe to the United States vs.

Traffic from the United States to Europe (Source: 2007 T100 International Market Data).

0 1 2 3 4 5 6 7 8 9

U.K.Germany

FranceNetherlands

ItalyIrelandSpain

SwitzerlandBelgium

Passengers between U.S. and Europe (Millions)

From U.S. to EuropeFrom Europe to U.S.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

JFK:LHRORD:LHRLAX:LHRJFK:CDGIAD:LHRSFO:LHRBOS:LHRMIA:LHRORD:FRAIAD:FRA

DTW:AMSEWR:LHRJFK:FRA

MCO:LGWSFO:FRA

MSP:AMSIAD:CDGATL:CDGLAX:CDGJFK:FCO

EWR:CDGJFK:AMSJFK:DUBIAH:LGWJFK:MADMIA:CDGEWR:FRAIAH:CDGDTW:FRAATL:FRA

Passengers for Top 30 Transatlantic Airport Pairs (Millions)

Europe to U.S.U.S. to Europe

81

Figure 4-6: 1990 – 2007 Historical Passenger Traffic from the United States to Denmark


A last factor in selecting the domain of analysis for the forecast model is that the

historical passenger traffic from the U.S. to Denmark, the country with the tenth most

passenger traffic in 2007, was abnormal. The historical passenger traffic for Denmark is

presented in Figure 4-6. The historical data did not lend itself to the credible forecast

modeling using the semi-log linear model eventually judged as preferred for the nine

countries with more passenger traffic. Once it was determined that historical passenger

traffic data for Denmark did not allow Denmark passenger traffic to be credibly modeled,

it was deemed inappropriate to include other European countries with passenger traffic

less than Denmark in 2007 in the domain of analysis.

4.2.3 Airports Included in Domain of Analysis

Thirty-one United States gateway airports and 35 European gateway airports are included

in the modeling domain of analysis. Gateway airports are defined as the airport serving

direct flights (either non-stop flights involving no intermediate stops or flights including a

stopover at an intermediate airport without a change in flight number) to another country.

The airports mentioned later in this report mean gateway airports.

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20071.5

2

2.5

3

3.5

4 x 105 10-Denmark (2007 Market Share: 1.5114) Yearly Passengers from U.S. 1990 - 2007

Year

Pass

enge

rs fr

om U

.S.

82

The 31 United States airports included in the domain of analysis were those with more

than 10,000 passengers to Europe per year in 2007. Figure 4-7 shows the 2007 passenger

traffic from each of selected 31 United States airports in the analysis domain to all the

selected 35 European airports. The other category includes the passenger traffic from

non-selected airports in the in the United States to whole Europe.

Figure 4-7: 2007 Passenger Traffic from Selected 31 United States Airports in the

analysis Domain to All the Selected 35 European Airports


0 1 2 3 4 5

JFK (New York, NY)EWR (Newark, NJ)ORD (Chicago, IL)

IAD (Washington, DC)ATL (Atlanta, GA)

LAX (Los Angeles, CA)BOS (Boston, MA)MIA (Miami, FL)

SFO (San Francisco, CA)PHL (Philadelphia, PA)

DTW (Detroit, MI)IAH (Houston, TX)MCO (Orlando, FL)DFW (Dallas, TX)

MSP (Minneapolis, MN)DEN (Seattle, WA)SEA (Denver, CO)

LAS (Las Vegas, NV)CLT (Charlotte, NC)

CVG (Cincinnati, OH)PHX (Baltimore, MD)MEM (Phoenix, AZ)PDX (Memphis, TN)TPA (Portland, OR)BWI (Sanford, FL)RDU (Tampa, FL)

RSW (Raleigh, NC)SFB (Ft. Myers, FL)SJU (San Juan, PR)BDL (Hartford, CT)

CLE (Cleveland, OH)Others

Passengers to Europe (Millions)

Included 31 U.S. airports

83

Figure 4-8: 2007 Passenger Traffic from All the Selected 31 United States Airports to

each 35 Selected European Airports in the analysis Domain


Thirty-five European airports from nine countries included in the geographical domain

are included in the domain of analysis. In addition, any other airports within the nine

selected geographical domain countries which had more than 10,000 passengers from the

United States in 2007 were also included in the domain of analysis. Figure 4-8 shows the

2007 passenger traffic from all the selected 31 United States airports in the analysis

domain to each of the selected 35 European airports. The other category includes the

passenger traffic from the United States to the non-selected European airports.

0 1 2 3 4 5 6

LHR (London, United Kingdom)LGW (London, United Kingdom)

MAN (Manchester, United Kingdom)GLA (Glasgow, United Kingdom)EDI (Edinburgh, United Kingdom)STN (London, United Kingdom)BFS (Belfast, United Kingdom)

BHX (Birmingham, United Kingdom)BRS (Bristol, United Kingdom)LTN (London, United Kingdom)

FRA (Frankfurt, Germany)MUC (Munich, Germany)

DUS (Dusseldorf, Germany)TXL (Berlin, Germany)

HAM (Hamburg, Germany)STR (Stuttgart, Germany)

CGN (Koeln, Germany)CDG (Paris, France)NCE (Nice, France)ORY (Paris, France)

AMS (Amsterdam, Netherlands)FCO (Rome, Italy)MXP (Milan, Italy)VCE (Venice, Italy)

PSA (Pisa, Italy)NAP (Naples, Italy)BLQ (Bologna, Italy)PMO (Palermo, Italy)DUB (Dublin, Ireland)

SNN (Shannon, Ireland)MAD (Madrid, Spain)

BCN (Barcelona, Spain)ZRH (Zurich, Switzerland)

GVA (Geneva, Switzerland)BRU (Brussels, Belgium)

Others

Passengers from U.S. (Millions)

Included 35 European airports

84

As shown in Figure 4-9 and Figure 4-10, the total of 66 airports in the United States and

Europe accounts for 91% of all the passenger traffic between the United States and

Europe in 2007.

Figure 4-9: 2007 Selected 31 United States Airports’ Cumulative Percentage of Total

Passenger Traffic from the United States to Europe


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

JFK (New

York, NY)

EWR (New

ark, NJ)

ORD

(Chicago, IL)

IAD (W

ashington, DC)

ATL (A

tlanta, GA)

LAX (Los Angeles, C

A)

BOS (Boston, M

A)

MIA (M

iami, FL)

SFO (San

Francisco, CA

)PH

L (Philade

lphia, PA)

DTW

(Detroit, M

I)IAH (H

ouston

, TX)

MCO

(Orlando

, FL)

DFW

(Dallas, TX)

MSP

(Minne

apolis, MN)

DEN

(Seattle, W

A)

SEA (D

enver, CO)

LAS (Las Vegas, N

V)

CLT (Charlotte, N

C)CV

G (Cincinn

ati, OH)

PHX (Baltimore, M

D)

MEM

(Ph

oenix, AZ)

PDX (M

emph

is, TN

)TPA (P

ortland, OR)

BWI (Sanford, FL)

RDU (Tam

pa, FL)

RSW (R

aleigh, NC)

SFB (Ft. Myers, FL)

SJU (San

Juan, PR)

BDL (Hartford, CT)

CLE (Cleveland

, OH)

Others

Cumulative Pe

rcen

tage

of P

assengers

United States airports with the top 31 passenger traffic in 2007 captured 91% of total passenger traffic from U.S. to Europe

91.2%

85

Figure 4-10: 2007 Selected 35 European Airports’ Cumulative Percentage of Total

Passenger Traffic from the United States to Europe


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

LHR (Lon

don, United Kingdo

m)

LGW (Lon

don, United Kingdo

m)

MAN (M

anchester, United Kingdo

m)

GLA

(Glasgow

, United Kingdo

m)

EDI (Edinbu

rgh, U

nited Kingdo

m)

STN (Lon

don, United Kingdo

m)

BFS (Belfast, United Kingdo

m)

BHX (Birmingham

, United Kingdo

m)

BRS (Bristol, United Kingdo

m)

LTN (Lon

don, United Kingdo

m)

FRA (Frankfurt, G

ermany)

MUC (M

unich, G

ermany)

DUS (Dusseldorf, Germany)

TXL (Berlin, G

ermany)

HAM (H

ambu

rg, Germany)

STR (Stuttgart, G

ermany)

CGN (Koe

ln, Germany)

CDG (P

aris, France)

NCE

(Nice, France)

ORY

(Paris, France)

AMS (Amsterdam, N

ethe

rlands)

FCO (Rom

e, Italy)

MXP

(Milan, Italy)

VCE

(Ven

ice, Italy)

PSA (Pisa, Italy)

NAP (Naples, Italy)

BLQ (Bologna, Italy)

PMO (P

alermo, Italy)

DUB (Dub

lin, Ireland)

SNN (Shann

on, Ireland)

MAD (M

adrid, Spain)

BCN (Barcelona, Spain)

ZRH (Zurich, Switzerland)

GVA (G

eneva, Switzerland)

BRU (B

russels, Belgium

)Others

Cumulative Pe

rcen

tage

of P

assengers

91.2%

Selected 31 European airports captured 91% of total passenger traffic from U.S. to Europe in 2007

86

4.3 Modeling Demand among the United States and Selected European Countries

4.3.1 Models Recommended

It is recommended that commercial air passenger traffic between the United States and

the nine European countries be modeled as a function of product of the U.S. Gross

Domestic Product (GDP) and the specific European country’s GDP with a dummy

variable to account for effect of September 11, 2001. Further, a semi-log relationship is

recommended. The estimated model is presented in Equation 4-1. As shown in Ln

α Ln _911 Equation 4-2, it was linearised through

logarithms:

exp α _911 Equation 4-1

Ln α Ln _911 Equation 4-2

where:

= Total passengers on the commercial flights from the U.S. to

European country j

= Product of European country j’s GDP and U.S. GDP in year 2000

millions of dollars

_911 = Dummy variable explaining 911 effect on traffic (0 for 1990 to

2000 and 1 for 2001-2007)

α = Estimated intercept

, = Estimated coefficients for European country j model

= Random error associated with model for European country j

The observed passenger demand from the U.S. to each of nine European countries from

the T100I Market data and the GDP (in millions of year 2000 dollars) from USDA

International Macroeconomic Data Set for the period 1990 – 2007 were used for the

calibration. The estimated coefficients and their statistical metrics are provided in Table

4-1. The R-squared values for nine estimated models range from 0.6 to 0.97. All

coefficients are of expected sign. The estimated coefficients are positive for the products

of the Gross Domestic Products (PGDP) and negative for Dummy_911. Except that the

coefficients for Dummy_911 for France and Ireland are significant at a ten percent (10%)

87

level on a one-tailed test, all the other estimated coefficients are five percent (5%) level

on a two-tailed test.

To further validate the passenger demand models, the historical passenger traffic from the

U.S. to each of nine European countries were compared with the corresponding estimated

passengers using the semi-log linear models. Figure 4-11 shows the comparison between

the historical U.S. to United Kingdom passenger traffic to the forecast. From Figure 4-11,

it is seen that the forecast follows the historical trend most of time though some

discrepancy exist between the historical value and the forecast in year 1991 and 2001.

The discrepancy in year 1991 was caused by the first gulf war, while the discrepancy in

year 2001 was caused by the 911 event. The comparisons for all the other eight countries’

models are included in Appendix C.3.

Table 4-1: Statistical Metrics for Semi-Log Country-to-Country Passenger Demand

Model.

Cj (t-value) β1j (t-value) β2j (t-value)1 U.K. -15.722 (-4.95) 1.046 (9.86) -0.249 (-3.81) 0.902 Germany -16.693 (-8.06) 1.042 (15.24) -0.135 (-4.06) 0.963 France -18.669 (-5.78) 1.108 (10.26) -0.098 (-1.67)** 0.944 Netherlands -43.190 (-11.19) 2.002 (14.83) -0.395 (-5.02) 0.965 Italy -26.339 (-5.68) 1.349 (8.64) -0.280 (-3.72) 0.876 Ireland -15.538 (-8.94) 1.052 (16.32) -0.118 (-1.72)** 0.987 Spain -13.060 (-4.71) 0.909 (9.50) -0.273 (-4.21) 0.908 Switzerland -22.655 (-3.69) 1.279 (5.88) -0.442 (-4.32) 0.679 Belgium -30.225 (-3.23) 1.535 (4.62) -0.870 (-4.79) 0.57*: 18 time-series observations during 1990 - 2007 are used for each country's regression analysis**: Significant at 0.10 level on a one-tailed test

j European CountrySemi-log Model*: Ln(Pj) = Cj + β1jLn(PGDPj) + β2jDummy_911 + ε

Estimated CoefficientsAdjusted R2

88

Figure 4-11: Comparison between Historical and Forecast Enplanements from the U.S. to

U.K. during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market

Data).

All the nine estimated models were applied using the forecast GDP (Appendix C.4:

Tables C-3 and C-4) of U.S. and each of nine European countries from USDA

International Macroeconomic Data Set for the period 2008 – 2020. The resulting forecast

passenger traffic was then adjusted by the adjustment factors. The adjust factor for each

of nine models is the difference between the base year (2007) historical passenger traffic

from the U.S. to the specific European country and the base year forecast. The adjust

factors for all the nine country-to-country passenger demand models are listed in Table

4-2. The resulting forecast passenger traffic from the U.S. to United Kingdom is

displayed in Figure 4-12. The forecast passenger traffic from the U.S. to each of the other

eight European countries is included in Appendix C.2.

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20074

4.5

5

5.5

6

6.5

7

7.5

8

8.5

9 x 106 1-United Kingdom (2007 Market Share: 33.44%) Yearly Passengers from U.S.

Year

Pass

enge

rs fr

om U

.S.

Forecast - Semi-log Model (R2 = 0.9)Historical

89

Table 4-2: Adjustment Factors for Semi-Logarithmic Country-to-Country Passenger

Demand Model.

Figure 4-12: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the

U.S. to United Kingdom (Historical Source: 1990 - 2007 T100 International Market

Data).

The product of U.S. population and destination European country population, and airfare

were also tested as variables which could affect passenger traffic. Surprisingly, airfare

was not found to be a significant contributor during the period of analysis (1990-2007).

The product of population was found to be significantly correlated with product of GDP

during the period of analysis (1990-2007). A description of other models examined is

contained next.

European Country Model Adjustment Factor1 United Kingdom -833,7492 Germany 346,7643 France -287,2224 Netherlands -408,5965 Italy 52,3336 Ireland 18,9817 Spain 108,1448 Switzerland -88,2149 Belgium -37,903

1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 20200.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2 x 107 1-United Kingdom (2007 Market Share: 33.44%) Yearly Passengers

Year

Pass

enge

rs fr

om U

.S.


90

4.3.2 Other Models Investigated

To forecast how passenger behavior and airline behavior would change with open skies,

the publically available data was examined along with the DB1B report for international

travel which was made available by the FAA. The data examined for forecasting

passenger traffic are:

- T100 transatlantic passenger survey (100% sample) from 1990 – 2007 (Appendix

C.4: Tables C-5 and C-6)

- The GDP data for US and European countries from USDA for the years 1990 through

2007 (Appendix C.4: Tables C-7 and C-8)

- The population data for US and European countries from USDA for the years 1990

through 2007 (Appendix C.4: Tables C-9)

- The DB1B airfare data (10% sample) for US to Europe travel from 1998 through

2007. This data allowed computation of US to European country average airfares for

each of the 9 years (Appendix C.4: Table C-10)

These data allow examining the following as independent variables affecting passenger

traffic between the US and Europe:

- GDP,

- Population,

- Average airfare, and

- The effect of September 11, 2001 attack was also examined.

Semi-logarithmic model was examined hypothesizing that passengers flying from the

United States to Europe were a function of population, the gross domestic products of the

United States and the destination European country, and the annual average airfare from

the United States to the European country and the effect for September 11, 2001.

A Fixed Effect Model was examined wherein the airfare elasticity was assumed to be

constant across all US-European country pairs while country pair specific intercept was

allowed for each US-European country pairs. The difference between country pairs is

represented by the country pair specific intercept. The model is presented in Equation 4-3

and Equation 4-4. Population was excluded in the model because the natural logarithms

91

of product of GDP (PGDP) was found to be highly correlated (correlation coefficient is

0.7) with the natural logarithms of product of population (PPopulation).

exp _911 Equation 4-3

Ln Ln _911 Equation 4-4

where:

= Total passengers on the commercial flights from the U.S. to European

country j

= Product of European country j’s GDP and U.S. GDP in year 2000

millions of dollars

Average airfare from the United States to European country j in year

2000 dollars

_911 Dummy variable explaining 911 effect on traffic (0 for 1990 to 2000 and

1 for 2001-2007)

α Estimated country-pair specific intercept

, , Estimated coefficients

Random error

To justify the assumed logarithmic linear relationship between passenger traffic and the

product of GDP (PGDP), the scatter plot of these two variables was made, as shown in

Figure 4-13. The apparent linear trend is seen from this figure. This relationship will be

justified further by the significant estimated coefficient for PGDP.

92

Figure 4-13: 1998 – 2007 Scatter Plot of Nature Logarithm of Passenger Traffic from the

United States to Selected Nine European Countries and Nature Logarithm of Product of

United States’ GDP and European Country’s GDP

(Passenger Traffic Source: 1998 - 2007 T100 International Market Data;

GDP Source: 1998 – 2007 USDA International Macroeconomic Data Set).

To justify the assumed logarithmic linear relationship between passenger traffic and the

real average airfare, the scatter plot of these two variables was made, as shown in Figure

4-14. Surprisingly, No apparent trend can be seen from this figure. Airfare was approved

not be a significant factor to affect the passenger traffic by the insignificant estimated

coefficient for airfare.

12

12.5

13

13.5

14

14.5

15

15.5

16

16.5

27 27.5 28 28.5 29 29.5 30 30.5 31

Ln(Passeng

ers)

Ln(Product of GDP)

93

Figure 4-14: 1998 – 2007 Scatter Plot of Nature Logarithm of Passenger Traffic from the

United States to Selected Nine European Countries and Nature Logarithm of Associated

Average Airfare from United States to Nine European Countries (Passenger Traffic

Source: 1998 - 2007 T100 International Market Data; Airfare Source: 1998 – 2007

DB1B Data).

The model was estimated by the Weighted Least Squares Regression approach using the

time-series (1998 – 2007) cross-section (nine countries) data. Each country was weighted

by its 2007 passenger traffic from the United States.

Table 4-3: Estimated Coefficients for the Fixed Effect Model Using Weighted Least

Squares Regression.

12

12.5

13

13.5

14

14.5

15

15.5

16

16.5

5.9 6 6.1 6.2 6.3 6.4 6.5 6.6

Ln(Passeng

ers)

Ln(Airfare)

Parameter B Std. Error t Sig.

95% Confidence Interval

Lower Bound Upper Bound

Intercept -5.383 3.350 -1.607 .112 -12.055 1.289Ln_PGDP .671 .125 5.384 .000 .423 .919Ln_Real_Airfare -.084 .161 -.523 .602 -.405 .236Dummy_911 -.155 .039 -4.008 .000 -.232 -.078[Country_ID=1] 1.530 .249 6.153 .000 1.035 2.025[Country_ID=2] .605 .273 2.220 .029 .062 1.148[Country_ID=3] .542 .241 2.249 .027 .062 1.023[Country_ID=4] 1.153 .099 11.683 .000 .956 1.349[Country_ID=5] -.140 .216 -.649 .518 -.571 .290[Country_ID=6] 11.819 2.095 5.642 .000 7.647 15.991[Country_ID=7] -.215 .160 -1.341 .184 -.533 .104[Country_ID=8] .480 .088 5.441 .000 .305 .656[Country_ID=9] 0a . . . . .a. This parameter is set to zero because it is redundant.

94

The estimated adjusted R-squared value for this model is 0.986. Equation 4-3 gives all

the estimated coefficients and the statistical metrics. As can been seen, the coefficients

for product of GDP and 911 attack dummy variable are statistically significant, while the

coefficient for airfare (underscored) is not statistically significant. This result was

unexpected. This is abnormal economic behavior as it is expected that commercial air

travel demand would decrease with increasing average airfare and that average airfare

would be a rather strong contributor to estimating passenger traffic.

The underlying reasons for this abnormal economic are perhaps in intuitively understood

by viewing the average airfare over time (Figure 4-15) and the annual passenger traffic

between US and Europe over time (Figure 4-16). From Figure 4-15 and Figure 4-16, it is

qualitatively observed that passenger traffic increase over time at the same time that

airfare increases. These figures show that air travel was not elastic with respect to airfares

during the analysis period. Understanding this abnormal economic behavior is a subject

for future research.

Figure 4-15: 1998 – 2007 Average Airfare from the United States to Selected Nine

European Countries (Data Source: 1998 - 2007 DB1B Data).

350

400

450

500

550

600

650

700

750

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Average

Airfare: 2000 Year $

Year

United KingdomGermany

France

Netherlands

Italy

Ireland

Spain

Switzerland

Belgium

95

Figure 4-16: 1998 – 2007 Passenger Traffic from the United States to Selected Nine

European Countries (Data Source: 1998 - 2007 T100 International Market Data).

With the finding that airfare is not a significant variable to forecast the passenger traffic,

the semi-logarithmic formulation recommended in section 4 was examined by using more

years (1990 – 2007) time-series data. It was found satisfactory and accepted as preferred.

4.4 Distribution of United States to European Country Passenger Demand to

Airport Pairs

To distribute the estimated future passenger traffic from the United States to each

European country to specific airport pairs, a Fratar (growth factor) model is used and an

iterative process is followed for each forecast year. Growth of airport pair’s passenger

traffic is projected from the base year, 2007 (2007 T100I Market Data).

Figure 4-17 shows simple spreadsheet example of the Fratar method and the iterative

process for one forecast year which will be described in parallel with the formulas for

0

1

2

3

4

5

6

7

8

9

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Passen

gers from

U.S. (Millions)

Year

United KingdomGermany

France

Netherlands

Italy

Ireland

Spain

Switzerland

Belgium

96

applying the growth factors and achieving convergence in the iterative process. Known at

the beginning of the distribution process from the T100I Market data are the base year

passengers (from US gateway airport oi to European gateway airport dj) (Figure 4-17, ,

shown in red type face in Iteration 0 , rows 3-6 and columns C-F), and the forecast

country-to-country passengers estimated by the socio-economic model described in

section 4 ( , (shown in red type face, row 9). To be determined are the annual

passengers for each airport pair in the forecast year (shown in blue type face for iterations

1, 2 and 3).

The process of distributing estimated passenger traffic may be thought of as filling

specific numbers in a spreadsheet where the spreadsheet row and column totals are

known and the individual numbers summing to the row and column totals are not known.

Iterative estimates of a forecast year passengers for each airport pair are made until the

total estimated enplaning passengers (from the United States) (Figure 4-17, blue type face

totals, column G), and deplaning passengers (in Europe) for each airport (Figure 4-17

blue type face, rows 17, 27, and 37 for iteration 1, 2, and 3, respectively) match the

forecast totals (Figure 4-17 green type face totals, column H and rows 18, 28, and 38 for

iteration 1, 2, and 3, respectively) within a desired tolerance.

Assumption: It is assumed that each United States airport in the domain of analysis will

retain its base year (2007) market share of total passenger traffic from the United States

to the nine European countries in the domain of analysis during the forecast period

(through 2020).

Total passengers from each United States airport in the forecast year are established using

Figure 4-17. These (Figure 4-17, column H) serve as one set target values for the iterative

estimates.

∑∑ ∑

Equation 4-5

Where:

is United States airport i’s estimated passengers to all (35) European airports

in forecast year t

97

is the estimated passengers from the United States to European country l in

forecast year t

is the passengers from United States airport i to European airport j in the base

year b, 2007 (2007 T100I Market Data)

Assumption: It is assumed that each European airport in the domain of analysis will

retain its base year (2007) market share of total passenger traffic from the United States

to its nation during the forecast period.

The total deplaning passengers to each European airport from all United States airports

are established per Figure 4-17. These (Figure 4-17 row 8 for iteration 0, row 18 for

iteration 1, etc.) serve as the other set of target values for the iterative estimates.

∑

∑ ∑ Equation 4-6

where:

is European airport j’s passengers for country l from all (31) United States

airports in year t

is the estimated passengers from the United States to European country l in

forecast year t

is the passengers from United States airport i to European airport j in the base

year b, 2007 (2007 T100I Market Data)

The iterative process is used to modify the transatlantic airport pair passenger demand in

the base year so that the total enplaning passengers at Unite States airports and deplaning

passengers at European airports match the forecasts within a desired tolerance.

Iterations are accomplished as follows: Adjustment factors, (Figure 4-17, in black type

face, column H and rows 10, 20, 30, and 40 for iteration 0, 1, 2, 3) are calculated from the

prior iteration data, and used to obtain the current iterative estimates of the passenger

traffic, , , between any Unite States airport and European airport pair using Equation

4-7. The adjustment factors for a specific iteration are the ratios of the forecast total

98

passengers to the total passengers for an airport estimated by the previously completed

iteration.

, , , , 12

∑ ,

∑ , , ∑ ,

∑ , ,

Equation 4-7

where: , is the estimated number of air passengers in the current (kth ) iteration from

United States airport i to European airport j , for the year t , is the estimated number of air passengers from United States airport i to

European airport j , for the previous (k-1)th iteration. For the first iteration, this

term is set at the base year (b) passengers, ,

, is adjustment factor for the U.S. airport i at )1( −k th iteration for year t

,∑ , ; ,

∑

, is adjustment factor for European airport j at )1( −k th iteration for year t

,∑ , ; ,

∑

Equation 4-7 may be expressed as follows. Passengers for kth iteration in forecast year t

between the United States airport i and European airport j are estimated to be the product

of four terms:

• the passengers estimated between the United States airport i and European airport j

for the previous ((k-1)th) iteration

• the adjustment factor for the total passengers from United States airport i to all 35

European airports

• the adjustment factor for the total passengers from all United States airports to

European airport j and

• the average of two terms:

99

o the inverse of the average of European airports’ adjustment factors (Figure

4-17, rows 10, 20, and 30) weighted by its estimated deplaning passenger

traffic (Figure 4-17, rows 17, 27, and 37) from United States airport i

o the inverse of the average of United States airports’ adjustment factors (Figure

4-17, column H) weighted by its estimated enplaning passenger traffic (Figure

4-17, column G) to European airport j

100

Figure 4-17: Spreadsheet illustration of Fratar Method and Iterative Process for

Distributing United States to Europe Passengers to Airport Pairs in Future Years.

A B C D E F G H I1

2 Iteration 0 oi\dj d1 d2 d3 d4 ∑jTijb,0 Oi

t Rit,0

3 o1 120 90 100 80 390 455 1.1666667

4 o2 100 85 90 70 345 403 1.1681159

5 o3 90 100 85 80 355 414 1.1661972

6 o4 100 85 90 75 350 408 1.1657143

7 ∑iTijb,0 410 360 365 305 1,440

8 Djt 490 430 414 346

9 Plt 1,680

10 Rjt,0 1.1948052 1.1948052 1.1343284 1.1343284

11

12 Iteration 1 oi\dj D1 D2 D3 D4 ∑jTijt,1 Oi

t Rit,1

13 o1 143.36162 107.52265 113.42120 90.73844 455 455 0.9999035

14 o2 119.62333 101.68119 102.21179 79.49935 403 403 0.9999611

15 o3 107.48700 119.43159 96.37734 90.70956 414 414 0.9999867

16 o4 119.40109 101.49229 102.02190 85.01963 408 408 1.0001595

17 ∑iTijt,1 490 430 414 346 1,680

18 Djt 490 430 414 346 1,680

19 Plt 1,680

20 Rjt,1 0.9999941 1.0000050 0.9999942 1.0000091

21

22 Iteration 2 oi\dj D1 D2 D3 D4 ∑jTijb,2 Oi

t Rit,2

23 o1 143.34708 107.51279 113.40961 90.73046 455 455 1.0000001

24 o2 119.61809 101.67772 102.20723 79.49694 403 403 1.0000001

25 o3 107.48500 119.43053 96.37547 90.70909 414 414 0.9999998

26 o4 119.41954 101.50895 102.03758 85.03391 408 408 1.0000000

27 ∑iTijt,2 490 430 414 346 1,680

28 Djt 490 430 414 346 1,680

29 Plt 1,680

30 Rjt,2 1.0000009 0.9999997 0.9999999 0.9999993

31

32 Iteration 3 oi\dj D1 D2 D3 D4 ∑jTijb,3 Oi

t Rit,3

33 o1 143.34722 107.51277 113.40961 90.73040 455 455 1.0000000

34 o2 119.61820 101.67770 102.20722 79.49688 403 403 1.0000000

35 o3 107.48507 119.43047 96.37544 90.70901 414 414 1.0000000

36 o4 119.41965 101.50893 102.03757 85.03385 408 408 1.0000000

37 ∑iTijt,3 490 430 414 346 1,680

38 Djt 490 430 414 346 1,680

39 Plt 1,680

40 Rjt,3 1.0000000 1.0000000 1.0000000 1.0000000

920 760

920 760

920 760

920 760

101

Iterations are performed until the totals of iterative based enplaning passengers at United

States airports (Figure 4-17 column G, ) are within one one-hundredth percent (0.01%) of

the forecasts for that year (Figure 4-17, column H ) and the totals of the deplaning

passengers at European airports (Figure 4-17, rows 17, 27, and 37) are within one one-

hundredth percent (0.01%) for the forecasts (Figure 4-17 rows 18, 28, and 38) . This is

accomplished by iterating until the adjustment factors converge to between 0.9999 and

1.0001. This convergence was usually reached within nine iterations for the 31 by 35

matrix (31 United States airports and 35 European airports in the domain of analysis). A

stop criterion for these iterations, set at 1000 iterations, was never approached. For the

spreadsheet example in Figure 4-17 a tighter adjustment factor convergence criteria of

1.0000000 was used in order to illustrate multiple iterations. In the simple spreadsheet

Fratar model example, convergence was essentially reached in the first iteration.

4.5 New Nonstop Flights Suggested by Forecast

68 new nonstop flights between the United States airports and the European airports are

forecast by the model in 2020 using the airport pair passenger demand forecast. These

new nonstop flights are forecast for two types of airport pairs: (1) pairs which in 2007

could only be travelled with connecting flights (2) and pairs which in 2007 could be

travelled in one flight which includes intermediate stops. 30 new nonstops flights are

forecast from 2010 on; 46 in 2015, and 68 in 2020. Table 4-4 shows the new nonstop

transatlantic airport pairs forecast in our analysis.

102

Table 4-4: New Nonstop Transatlantic Airport Pairs and Their Forecast Passenger

Demand in Year 2010, 2015 and 2020.

US Airport

US City, StateEuropean Airport

European City, County Passengers Year Remark*

1 LAS Las Vegas, NV LHR London, United Kingdom 128,913 2007 02 LAX Los Angeles, CA FCO Rome, Italy 102,856 2007 13 MCO Orlando, FL LHR London, United Kingdom 92,857 2007 04 DFW Dallas/Ft.Worth, TX LHR London, United Kingdom 77,596 2007 15 SFO San Francisco, CA FCO Rome, Italy 76,624 2007 16 LAX Los Angeles, CA MAD Madrid, Spain 70,011 2007 17 SEA Seattle, WA FRA Frankfurt, Germany 62,464 2007 18 SFO San Francisco, CA MAD Madrid, Spain 59,574 2007 09 MCO Orlando, FL CDG Paris, France 56,635 2007 110 LAS Las Vegas, NV AMS Amsterdam, Netherlands 56,017 2007 111 PHX Phoenix, AZ FRA Frankfurt, Germany 53,784 2007 112 MSP Minneapolis/St. Paul, MN FRA Frankfurt, Germany 51,702 2007 013 LAS Las Vegas, NV CDG Paris, France 48,573 2007 114 DEN Denver, CO CDG Paris, France 45,900 2007 015 DEN Denver, CO AMS Amsterdam, Netherlands 65,480 2010 116 DTW Detroit, MI FCO Rome, Italy 53,252 2010 017 SFO San Francisco, CA MAN Manchester, United Kingdom 50,434 2010 118 MSP Minneapolis/St. Paul, MN CDG Paris, France 49,748 2010 019 TPA Tampa, FL LHR London, United Kingdom 48,987 2010 020 LAX Los Angeles, CA BCN Barcelona, Spain 48,402 2010 121 LAX Los Angeles, CA MAN Manchester, United Kingdom 48,075 2010 122 LAX Los Angeles, CA MXP Milan, Italy 47,518 2010 123 TPA Tampa, FL CDG Paris, France 47,317 2010 024 PDX Portland, OR LHR London, United Kingdom 46,870 2010 125 MIA Miami, FL FCO Rome, Italy 46,465 2010 126 RDU Raleigh/Durham, NC LHR London, United Kingdom 45,998 2010 027 PHX Phoenix, AZ CDG Paris, France 45,362 2010 028 TPA Tampa, FL FRA Frankfurt, Germany 45,083 2010 129 DFW Dallas/Ft.Worth, TX AMS Amsterdam, Netherlands 45,028 2010 030 PHX Phoenix, AZ AMS Amsterdam, Netherlands 44,108 2010 0

103

Table 4-4: New Nonstop Transatlantic Airport Pairs and Their Forecast Passenger

Demand in Year 2010, 2015 and 2020 (Continued).

4.6 Criteria for suggesting new nonstop flights

Analysis of 2007 airline behavior indicates that airlines are likely to establish 12 month

service for transatlantic airport pairs when the passenger demand per year is in the

neighborhood of 44,000. This translates to five or more flights per week.

US Airport

US City, StateEuropean Airport

European City, County Passengers Year Remark*

31 TPA Tampa, FL AMS Amsterdam, Netherlands 70,808 2015 132 LAX Los Angeles, CA BRU Brussels, Belgium 61,126 2015 133 DFW Dallas/Ft.Worth, TX FCO Rome, Italy 58,109 2015 034 LAX Los Angeles, CA LGW London, United Kingdom 57,417 2015 135 SFO San Francisco, CA BRU Brussels, Belgium 55,999 2015 036 PDX Portland, OR AMS Amsterdam, Netherlands 54,618 2015 037 SFO San Francisco, CA MXP Milan, Italy 53,259 2015 038 ORD Chicago, IL BCN Barcelona, Spain 52,021 2015 139 MSP Minneapolis/St. Paul, MN FCO Rome, Italy 51,958 2015 040 RDU Raleigh/Durham, NC CDG Paris, France 51,242 2015 041 MCO Orlando, FL MAD Madrid, Spain 49,487 2015 142 ATL Atlanta, GA LHR London, United Kingdom 48,514 2015 043 CLE Cleveland, OH AMS Amsterdam, Netherlands 47,322 2015 044 PDX Portland, OR CDG Paris, France 44,974 2015 045 DFW Dallas/Ft.Worth, TX MAD Madrid, Spain 44,254 2015 046 LAS Las Vegas, NV DUB Dublin, Ireland 44,167 2015 047 MSP Minneapolis/St. Paul, MN LHR London, United Kingdom 57,527 2020 048 SEA Seattle, WA FCO Rome, Italy 56,589 2020 049 RDU Raleigh/Durham, NC AMS Amsterdam, Netherlands 55,804 2020 050 MIA Miami, FL LGW London, United Kingdom 55,580 2020 051 DTW Detroit, MI MUC Munich, Germany 54,949 2020 152 MIA Miami, FL MAN Manchester, United Kingdom 54,946 2020 053 SFO San Francisco, CA BCN Barcelona, Spain 53,349 2020 054 LAX Los Angeles, CA VCE Venice, Italy 52,446 2020 155 MIA Miami, FL BRU Brussels, Belgium 52,430 2020 056 DEN Denver, CO FCO Rome, Italy 52,393 2020 057 RDU Raleigh/Durham, NC FRA Frankfurt, Germany 52,236 2020 058 CLT Charlotte, NC CDG Paris, France 52,067 2020 059 IAH Houston, TX FCO Rome, Italy 51,198 2020 160 MIA Miami, FL BCN Barcelona, Spain 51,032 2020 161 SFO San Francisco, CA LGW London, United Kingdom 50,988 2020 162 LAX Los Angeles, CA TXL Berlin, Germany 50,968 2020 163 CLE Cleveland, OH FRA Frankfurt, Germany 50,078 2020 064 ORD Chicago, IL LGW London, United Kingdom 49,606 2020 065 ORD Chicago, IL VCE Venice, Italy 48,600 2020 066 LAS Las Vegas, NV MUC Munich, Germany 48,558 2020 067 ORD Chicago, IL TXL Berlin, Germany 45,100 2020 068 CLE Cleveland, OH CDG Paris, France 45,094 2020 0

*: 1 ‐ Airport pairs only provided with service with intermediate stop0 ‐ Airport pairs only provided with connecting service

104

To understand when new nonstop flights between gateway airport pairs are likely to be

instituted by the airlines, the 2007 T100 International Segment data was examined to

infer the airlines’ threshold values of passenger demand and flight frequency to offer

service. It is judged that the annual passenger demand for a pair of gateway airports

served by only one air carrier is a credible indicator of the passenger demand threshold

for airlines to establish nonstop service between an airport pair.

Of the 1085 possible transatlantic airport pairs studied (31 United States airports and the

35 European airports in the domain of analysis), 214 pairs have nonstop flights and 59

pairs have flights with intermediate stops. Travel among the remaining 812 airport pairs

is accomplished by crossing the Atlantic on one of the 214 nonstop flights or one of the

59 flights with an intermediate stop(s) and connecting via other flights.

Figure 4-18 shows the number of airlines providing nonstop service to each of the 214

transatlantic airport pairs in the domain of analysis in 2007. 135 of the 214 airport pairs

are served by only one airline, while the remainders are served by two or more carriers.

Figure 4-18: Distribution of the number of airlines providing nonstop service

(Source: 2007 T100I Segment Data).

135

51

21

4 2 ‐ 1 ‐

20

40

60

80

100

120

140

160

1 2 3 4 5 6 7

No. of Transatlantic Airpo

rt‐Pairs

No. of Commercial Carriers Operating Non‐stop Flights

105

Year-round service is provided for 89 of the 135 pairs where nonstop service was

provided in 2007 by a single airline while 46 of airport pairs are served less than 12

months per year. Figure 4-19 shows the distribution of the number of months the nonstop

service was provided for 135 pairs by a single airline.

Forecasts for seasonal markets (with less than 12 month service) are not undertaken in

this analysis. The passenger demand models discussed in section 4 are annual passenger

demand models and are not deemed effective at seasonal effects in monthly passenger

demand.

Figure 4-19: Distribution of the number of months for single airline to provide nonstop

service (Source: 2007 T100I Segment Data).

4 2 2 4 2 10 8 5 3 3 3

89

‐

10

20

30

40

50

60

70

80

90

100

1 2 3 4 5 6 7 8 9 10 11 12

No. of Transatlantic Airpo

rt‐Pairs

No. of Months Served with Non‐stop Flights

106

Figure 4-20: Passengers per airport pair per year for the 89 airport pairs provided with

nonstop flight by a single airline (Source: 2007 T100I Segment Data).

Figure 4-20 shows the cumulative distribution of the 2007 T100I Segment annual

passenger demand for the 89 airport pairs in the domain served by one airline for 12

months. It is seen that 95% of the passengers traveling between these 89 pairs served by

one carrier have annual enplanements in the neighborhood of 44,000 or more. It should

be noted that 10 pairs with less than 44,000 passengers per year have nonstop service.

(One airport pair that has a very low number of annual passenger enplanement is Newark,

NJ, to Zurich, Switzerland, where business class service is an exclusive air service

offering). The 95% cutoff point encompasses 79 of the pairs served by one carrier.

Figure 4-21 shows the cumulative distribution of 2007 annual flights for the 89 airport

pairs in the domain served by one airline for 12 months. It can be seen that about 95

percent of the 89 airport pairs are served with 260 flights per year, which is five flights

per week.

0%5%10%15%20%25%30%35%40%45%50%55%60%65%70%75%80%85%90%95%

100%

7,428 57,428 107,428 157,428 207,428 257,428 307,428 357,428

Cumulative Pe

rcen

tage

of P

assengers

2007 Passengers per Airport Pair

44,415

107

Figure 4-21: Flight frequency per airport pair per year for the 89 airport pairs 89 airport

pairs provided with nonstop flight by a single airline (Source: 2007 T100I Segment

Data).

Assumption: Based on the foregoing analysis, it is assumed that the threshold for

airlines to provide nonstop service to an airport pair is in the neighborhood of 44,000

passengers per year.

With the identification of demand threshold for establishing nonstop service, demand is

forecast between all gateway airport pairs in the analysis time frame for comparison the

threshold value to identify potential new nonstop flights.

4.7 Demand for travel between gateways without nonstop flights

New nonstop flights are expected to be established when the estimated demand between

the 871 (812+59) pairs reaches the threshold observed for nonstop transatlantic service

(i.e., 44,000 passengers/year).

Travel between 871 airport pairs (not provided with nonstop service) is accomplished by

traveling across the Atlantic using one of gateway airport pairs with nonstop service.

Passenger demand for the 871 pairs that do not have nonstop service is estimated by

determining how many passengers for each of the 871 pairs travel between each of the

0%5%

10%15%20%25%30%35%40%45%50%55%60%65%70%75%80%85%90%95%100%

123 323 523 723 923 1,123 1,323 1,523

Cumulative Pe

rcen

tage

of Flights

2007 Flight Frequency per Airport Pair

260

108

182 gateway airport pairs that have credible direct passenger forecast resulted from

section 5. The analysis considers 98% of total passengers from the U.S. to selected nine

European countries in 2007.

2007 DB1B data is used to derive the fraction of demand for each of the 182 airport pairs

that is attributed to the 871 airport pairs (that are not provided nonstop flight service).

This fraction is assumed to remain constant during the forecast period. For 39 of the

182 gateway airport pairs, the DB1B sample was less than one percent of the T100I

Market survey. These pairs were not included in the new nonstop service analysis as they

are not considered credible samples. Future demand for the 871 pairs is then estimated

using the demand forecast for the 143 (=189-39) credible pairs and the fraction of this

demand attributed to each of the 871 pairs.

To illustrate this computation, consider the demand between Atlanta, Hartsfield and

London, Heathrow. All 2007 DB1B records are examined for each ATL - LHR candidate

gateway pair. For illustration purposes the records are grouped according to the airport

pairs used to cross the North Atlantic without stops. DB1B sample records show that

616 passengers traveling through Atlanta, Hartsfield (ATL) and London, Heathrow

(LHR) used Chicago, O’Hare (ORD) London. All records between ATL-LHR passing

through ORD include 104 unique routes. All these routes are shown in Figure 4-22.

Figure 4-22: 104 Unique Routes for Passengers Travelling through ATL and LHR using

ORD and LHR airport pair where nonstop service is provided (Source: 2007 DB1B

Data).

109

Overall DB1B sample records include 1,321 passengers using 41 of the 143 credible

gateway airport pairs for travel between Atlanta, and London, Heathrow. The fraction of

passengers traveling between ATL and LHR using the ORD-LHR gateway airport pair is

616/50,608. Combined with total passengers using ORD-LHR gateway airport pair in

2007 (754,060) and 2010 (991,612), it yields 9,178 passengers in 2007 and 12,070 for

2010.

Table 4-5 shows completion of the process for all 41 gateway airport pairs including

ORD-LHR, with 27,282 passengers being estimated for this pair in 2007 and 35,929

being forecast in 2010. The demand forecast results for all 871 airport pairs without

nonstop service are shown in Table 4-6 for 2010.

110

Table 4-5: Process for Estimating Total Passenger Travelling between Atlanta and

London, Heathrow.

ATL ‐ LHR Gateway

Airport Pairs

2007 ATL ‐ LHR DB1B Passengers

2007 No. of Routes

2007 All DB1B

Passengers

Fraction of ATL ‐ LHR Passengers

2007 T100I Passengers

2007 Passenger Demand for ATL ‐ LHR

2010 Forecast T100I

Passengers

2010 Passenger Demand for ATL ‐ LHR

(A) (B) (C) (D = A/C) (E) (F = E*D) (G) (H = G*D)1 ORD:LHR 616 104 50,608 1.217% 754,060 9,178 991,612 12,070 2 IAD:LHR 291 59 22,426 1.298% 498,183 6,464 650,727 8,444 3 JFK:LHR 123 53 35,955 0.342% 1,352,621 4,627 1,768,333 6,049 4 ATL:LGW 77 63 11,297 0.682% 190,248 1,297 259,781 1,771 5 MIA:LHR 38 15 6,807 0.558% 408,631 2,281 544,832 3,042 6 BOS:LHR 35 18 13,046 0.268% 424,873 1,140 545,148 1,463 7 ATL:CDG 30 14 15,239 0.197% 258,415 509 338,277 666 8 ATL:AMS 24 8 6,609 0.363% 135,832 493 227,441 826 9 ATL:FRA 15 6 10,478 0.143% 192,871 276 208,550 299 10 ATL:MAN 8 4 6,567 0.122% 70,532 86 93,741 114 11 ATL:DUB 7 5 5,104 0.137% 54,678 75 68,086 93 12 EWR:LGW 5 5 13,563 0.037% 145,812 54 196,106 72 13 ATL:EDI 5 4 3,431 0.146% 36,438 53 47,889 70 14 DTW:AMS 4 1 25,196 0.016% 366,228 58 544,606 86 15 ORD:MAN 4 2 11,759 0.034% 128,649 44 170,453 58 16 JFK:CDG 3 2 22,096 0.014% 598,834 81 776,914 105 17 ORD:DUB 3 1 8,698 0.034% 136,512 47 169,458 58 18 ATL:MXP 3 2 5,595 0.054% 66,608 36 76,442 41 19 ATL:MAD 3 3 5,790 0.052% 65,572 34 67,384 35 20 LAX:LHR 2 2 19,297 0.010% 720,705 75 919,287 95 21 JFK:LGW 2 2 7,438 0.027% 77,608 21 105,027 28 22 CVG:LGW 2 2 4,852 0.041% 52,583 22 71,855 30 23 ORD:SNN 2 1 3,264 0.061% 51,575 32 63,966 39 24 ATL:SNN 2 2 2,729 0.073% 28,096 21 34,955 26 25 IAH:LGW 1 1 11,608 0.009% 229,033 20 283,705 24 26 DFW:LGW 1 1 13,204 0.008% 189,793 14 258,458 20 27 BOS:CDG 1 1 4,117 0.024% 177,834 43 226,437 55 28 MIA:MAD 1 1 6,643 0.015% 172,784 26 179,466 27 29 ORD:AMS 1 1 6,472 0.015% 136,363 21 227,622 35 30 JFK:SNN 1 1 4,841 0.021% 123,275 25 151,993 31 31 ATL:FCO 1 1 9,487 0.011% 101,673 11 116,759 12 32 JFK:MAN 1 1 3,906 0.026% 93,862 24 123,636 32 33 MSP:LGW 1 1 8,295 0.012% 84,331 10 90,195 11 34 ATL:MUC 1 1 6,022 0.017% 66,962 11 71,506 12 35 EWR:GLA 1 1 6,655 0.015% 65,693 10 85,998 13 36 PHL:LGW 1 1 6,258 0.016% 62,962 10 87,872 14 37 ATL:STR 1 1 5,728 0.017% 59,686 10 63,934 11 38 ATL:BRU 1 1 5,179 0.019% 58,228 11 80,427 16 39 EWR:BHX 1 1 5,515 0.018% 54,816 10 71,276 13 40 ATL:DUS 1 1 4,202 0.024% 45,506 11 48,557 12 41 ATL:BCN 1 1 3,735 0.027% 41,070 11 42,579 11

Total 1,321 ‐ ‐ ‐ ‐ 27,282 ‐ 35,929

111

Table 4-6: 2010 Passenger Traffic for 871 Transatlantic Airport Pairs without Nonstop Service.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17LHR LGW MAN GLA EDI STN BFS BHX BRS LTN FRA MUC DUS TXL HAM STR CGN

1 JFK ‐ ‐ ‐ 14,438 16,461 ‐ 39 2,923 623 ‐ ‐ ‐ ‐ ‐ ‐ ‐ 1,734 2 EWR ‐ ‐ ‐ ‐ ‐ 42 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 2,533 ‐ 3 ORD ‐ 27,774 ‐ 18,217 20,373 489 899 3,576 1,034 ‐ ‐ ‐ ‐ 25,863 19,770 12,890 3,675 4 IAD ‐ 6,266 10,733 8,758 12,342 ‐ 676 1,868 483 ‐ ‐ ‐ 10,817 20,365 15,049 14,032 4,024 5 ATL 35,929 ‐ ‐ 1,955 ‐ 96 962 2,949 1,342 ‐ ‐ ‐ ‐ 14,365 7,063 ‐ 986 6 LAX ‐ 43,326 48,075 ‐ 19,712 ‐ 1,287 2,921 1,538 ‐ ‐ ‐ ‐ 28,378 19,871 12,140 5,541 7 BOS ‐ ‐ ‐ ‐ 14,002 522 1,639 3,581 2,186 ‐ ‐ ‐ 6,532 14,167 6,629 5,245 2,852 8 MIA ‐ 30,401 30,294 9,222 6,575 726 976 961 483 ‐ ‐ ‐ ‐ 10,815 7,724 6,460 606 9 SFO ‐ 29,410 50,434 13,813 17,765 458 1,779 3,292 1,500 ‐ ‐ ‐ 11,984 19,505 16,652 9,266 4,608 10 PHL ‐ ‐ ‐ ‐ 721 125 39 319 56 ‐ ‐ ‐ 3,461 9,026 4,563 4,729 467 11 DTW ‐ ‐ 14,725 6,996 6,862 12 672 9,729 2,868 ‐ ‐ 27,100 ‐ 12,821 8,308 18,210 4,691 12 IAH 23,032 ‐ 11,945 6,108 7,500 44 2,223 3,457 2,682 ‐ ‐ 8,645 2,616 5,398 5,002 2,120 3,411 13 MCO 121,638 ‐ ‐ ‐ 10,189 347 4,695 4,561 2,721 ‐ ‐ 24,462 7,828 9,003 4,913 3,726 918 14 DFW 102,035 ‐ 16,384 4,854 7,951 470 729 956 528 ‐ ‐ 16,705 2,336 4,987 3,330 2,962 747 15 MSP 31,787 ‐ 10,424 5,270 5,755 ‐ 1,229 4,320 1,188 ‐ 55,641 22,591 7,794 10,048 6,125 6,895 2,990 16 DEN ‐ 19,279 15,909 2,851 5,177 17 976 1,563 966 ‐ ‐ ‐ 3,906 7,175 6,229 4,186 1,745 17 SEA ‐ 18,591 9,281 2,413 4,220 197 937 1,683 901 ‐ 70,815 21,961 3,064 6,259 3,377 3,888 1,520 18 LAS 167,195 ‐ ‐ 13,591 9,551 ‐ 2,173 2,958 1,770 ‐ ‐ 29,868 ‐ 4,542 4,462 3,525 992 19 CLT 18,680 ‐ 9,719 1,796 2,100 29 611 1,063 468 ‐ ‐ ‐ 4,049 5,577 3,343 2,867 673 20 CVG 10,774 ‐ 4,965 1,181 2,379 ‐ 195 689 442 ‐ ‐ 7,396 1,919 3,896 2,216 2,237 147 21 PHX ‐ 23,162 10,343 1,784 2,220 57 455 659 600 ‐ 58,916 22,589 1,944 3,229 2,505 1,913 698 22 MEM 12,541 12,565 5,900 2,549 2,954 ‐ 338 2,345 937 ‐ 14,550 4,446 1,453 1,589 1,861 2,220 464 23 PDX 46,870 13,399 5,329 1,594 2,707 84 416 662 505 ‐ ‐ 22,009 1,061 3,241 1,914 1,921 552 24 TPA 48,987 ‐ 16,657 3,199 4,799 94 1,392 2,561 1,595 ‐ 45,083 15,589 1,953 5,059 4,114 3,623 676 25 BWI ‐ 5,837 5,119 592 1,019 62 325 403 520 ‐ 18,768 2,566 483 1,482 661 1,329 275 26 RDU 45,998 ‐ 6,085 2,320 3,211 243 702 574 645 ‐ 32,512 9,541 2,194 2,703 1,604 2,775 745 27 RSW 12,182 10,891 6,600 1,442 1,732 33 688 1,031 676 ‐ ‐ ‐ ‐ 1,565 1,804 2,159 150 28 SFB ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 29 SJU 1,952 57 193 68 ‐ ‐ 26 13 ‐ ‐ 254 14 122 22 ‐ ‐ ‐ 30 BDL 5,919 2,242 1,224 233 420 14 65 443 180 ‐ 7,595 2,247 404 637 430 339 147 31 CLE 18,024 ‐ 5,819 1,711 2,260 17 714 2,568 1,325 ‐ 31,546 7,871 1,476 2,470 1,907 1,231 1,757

Year = 2010 (Passenger Demand)

1 ‐ U.K. 2 ‐ Germany

112

Table 4-6: 2010 Passenger Traffic for 871 Transatlantic Airport Pairs without Nonstop Service (Continued).

4 ‐ Netherlands 9 ‐ Belgium18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35CDG NCE ORY AMS FCO MXP VCE PSA NAP BLQ PMO DUB SNN MAD BCN ZRH GVA BRU

1 JFK ‐ ‐ 3,009 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 2 EWR ‐ 5,503 ‐ ‐ ‐ ‐ 10,976 1,525 6,006 4,321 4,151 ‐ ‐ ‐ ‐ ‐ ‐ ‐ 3 ORD ‐ 16,685 868 ‐ ‐ ‐ 27,880 2,753 8,682 5,977 3,620 ‐ ‐ ‐ 39,197 ‐ 15,658 ‐ 4 IAD ‐ 12,665 366 ‐ ‐ 13,945 17,671 2,019 6,823 3,181 1,595 ‐ 4,805 ‐ 17,545 ‐ 13,448 ‐ 5 ATL ‐ 13,913 2,013 ‐ ‐ ‐ ‐ 3,115 6,484 3,858 3,551 ‐ ‐ ‐ ‐ ‐ 8,247 ‐ 6 LAX ‐ 14,893 344 ‐ 122,020 47,518 28,351 1,646 4,308 7,299 2,565 ‐ 14,449 75,606 48,402 ‐ 18,183 42,471 7 BOS ‐ 8,745 1,040 ‐ ‐ ‐ 8,054 1,735 1,974 2,877 378 ‐ ‐ ‐ 18,294 ‐ 8,654 20,007 8 MIA ‐ 11,961 1,857 ‐ 46,465 ‐ 8,914 1,074 1,940 2,808 617 22,734 6,365 ‐ 29,378 ‐ 5,267 25,538 9 SFO ‐ 12,699 348 ‐ 88,389 40,425 21,805 1,677 4,857 5,238 1,924 ‐ 20,502 62,665 31,873 ‐ 13,996 39,366 10 PHL ‐ 3,226 366 ‐ ‐ ‐ ‐ 1,201 5,543 1,510 4,296 ‐ ‐ ‐ ‐ ‐ 2,450 ‐ 11 DTW ‐ 6,365 352 ‐ 53,252 14,873 14,026 430 1,249 7,454 1,708 11,613 5,310 17,495 17,924 15,516 9,763 ‐ 12 IAH ‐ 4,331 848 ‐ 28,500 16,805 6,224 503 1,014 1,013 446 7,005 3,708 16,116 11,926 9,878 6,386 9,342 13 MCO 73,348 4,181 255 ‐ 23,637 11,887 5,359 370 1,041 699 604 ‐ 19,175 38,179 9,802 13,906 3,187 15,180 14 DFW ‐ 5,792 443 45,028 43,001 10,180 11,996 1,942 3,024 2,705 313 18,335 6,025 33,961 12,685 ‐ 4,214 14,391 15 MSP 49,748 4,346 39 ‐ 37,196 13,752 11,862 209 963 4,541 391 ‐ 7,377 17,935 12,615 17,198 9,262 12,011 16 DEN 58,436 3,320 92 65,480 29,896 10,171 9,439 697 2,280 1,469 251 12,730 5,202 13,624 8,820 17,795 5,499 10,998 17 SEA ‐ 4,269 129 ‐ 30,544 10,326 8,155 643 1,461 1,472 451 15,070 4,656 13,856 9,395 11,744 4,185 9,407 18 LAS 62,369 2,590 195 84,922 14,309 13,577 5,083 549 763 2,127 370 31,948 6,482 17,875 6,335 25,310 3,485 16,306 19 CLT 29,720 1,652 97 ‐ 15,315 6,171 6,102 398 1,505 1,099 115 6,250 2,624 5,710 4,645 5,270 1,820 4,590 20 CVG ‐ 4,780 473 ‐ ‐ 8,576 6,907 310 1,072 596 747 8,103 4,841 6,834 5,550 5,289 5,330 6,339 21 PHX 45,362 1,737 185 44,108 21,933 8,371 6,948 140 641 717 405 10,416 3,263 10,370 6,561 10,146 2,076 7,968 22 MEM 11,508 1,174 54 ‐ 8,722 2,526 3,241 51 419 854 47 3,527 1,037 3,303 3,398 2,620 1,833 3,495 23 PDX 34,283 1,450 97 35,327 16,049 5,196 4,094 497 856 1,241 238 9,833 2,498 5,715 5,836 8,649 2,295 4,470 24 TPA 47,317 2,543 44 43,972 21,508 8,132 5,157 320 1,183 548 658 12,096 6,022 11,756 6,122 8,877 1,906 7,688 25 BWI 14,491 759 57 8,998 9,534 2,270 2,519 197 367 98 292 10,587 4,925 4,049 2,774 2,743 888 2,922 26 RDU 38,783 1,851 133 22,023 15,306 4,664 3,650 576 1,036 744 258 7,574 2,759 9,020 3,783 10,735 2,195 6,464 27 RSW 10,838 366 16 9,633 4,043 1,301 981 11 69 124 47 4,514 1,429 1,512 771 3,904 768 2,007 28 SFB ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 29 SJU 277 ‐ ‐ 101 42 78 ‐ ‐ 35 ‐ ‐ 161 31 ‐ 22 339 119 77 30 BDL 5,324 210 21 ‐ 3,402 817 648 29 87 29 32 1,248 437 1,076 476 1,304 319 1,531 31 CLE 26,312 1,011 20 29,576 19,687 9,013 2,750 105 460 435 502 8,367 4,257 5,320 3,930 8,320 2,577 5,750

7 ‐ Spain 8 ‐ Switzerland3 ‐ France 5 ‐ Italy 6 ‐ IrelandYear = 2010 (Passenger Demand)

113

4.8 Demonstration of Capacity Constraints

As demand for air travel continues to grow in the future, passenger traffic at certain airports will

reach that airport’s passenger capacity. Part of the traffic at these airports will, of necessity, be

diverted to other airports or perhaps even choose other modes of transportation. The purpose of

this section is to demonstrate a method for rerouting the excess air travel passengers from one

airport to other airports when an airport’s passenger capacity is exceeded assuming all rerouted

travelers continue to use the commercial air travel mode. As done before due to symmetry,

traffic is analyzed from the United States to Europe. This method offered address passengers

from the United States only. To understand all the diversion of passenger traffic from London,

Heathrow, similar analyses would be performed for all other regions of the world.

The method is presented followed by an example calculation/analysis of reassigning excess

passenger traffic at London, Heathrow airport.

When the number of passengers served per year reaches a European airport’s passenger capacity,

it is assumed that the number of passengers from the United States to the “at-capacity” airport

will be held at the existing level when demand reached capacity. United States Passenger

demand beyond this level is termed excess and divided into two categories: passengers who

would normally terminate at the “at-capacity” airport and passengers who would normally

connect to other airports through the “at-capacity” airport:

1. The excess terminating passengers from the United States are reassigned to selected nearby

candidate airports which already handle similar terminating passengers. It is assumed that the

candidate airports have sufficient capacity to process the rerouted passengers. The specific

number of excess passengers reassigned to each candidate nearby airport is proportionate to

the candidate airport’s fraction of the total number of T100 Market passengers from the

United States for all the candidate airports. In the event, there are no nearby airports which

handle terminating passengers, then all excess passengers are assigned to the excess

connecting passenger category and rerouted in that manner.

2. The excess connecting passengers from the United States are rerouted to connect at other

European candidate airports which are frequently used for connecting. It is assumed that the

candidate airports have sufficient capacity to process the rerouted passengers. The specific

114

number of excess passengers reassigned to each candidate nearby airport is proportionate to

the candidate airport’s fraction of the total number of connecting passengers from the United

States at all the candidate airports. The number of United States connecting passengers at

each airport is calculated as the product of the DB1B connecting passenger ratio and the base

year (T100I Market) passenger demand. The DB1B connecting passenger ratio is the ratio of

the DB1B sample of connecting passengers to the sum of the DB1B samples of connecting

and terminating passengers.

115

Table 4-7: Estimated DB1B Connecting Passenger Ratio at 35 European Airports in Domain in

Base Year (2007) (Source: 2007 DB1B Data).

Eighty percent (80%) of passengers from the United States connected at 13 European airports in

the base year (2007). The DB1B connecting passenger ratio at these airports ranges from 49% to

94%.

European Airport

2007 DB1B Terminating Passengers

2007 DB1B Connecting Passengers

2007 DB1B Passengers from U.S.

2007 DB1B Connecting

Passenger Ratio

CDF of 2007 DB1B Connecting Passengers

(A) (B) (C = A+B) (D = B/C) (E)1 LHR 54,275 124,198 178,473 69.6% 12.8%2 CDG 37,624 101,326 138,950 72.9% 23.3%3 FRA 62,713 92,737 155,450 59.7% 32.9%4 LGW 15,613 91,203 106,816 85.4% 42.3%5 FCO 9,087 73,550 82,637 89.0% 49.9%6 AMS 66,026 63,102 129,128 48.9% 56.4%7 MAN 3,101 45,518 48,619 93.6% 61.1%8 MAD 12,955 37,131 50,086 74.1% 65.0%9 DUB 3,846 35,932 39,778 90.3% 68.7%

10 BRU 6,627 34,613 41,240 83.9% 72.3%11 MUC 22,526 32,220 54,746 58.9% 75.6%12 ZRH 7,448 27,258 34,706 78.5% 78.4%13 BCN 3,126 27,256 30,382 89.7% 81.2%14 MXP 9,451 21,796 31,247 69.8% 83.5%15 SNN 3,001 17,237 20,238 85.2% 85.3%16 TXL 1,340 15,701 17,041 92.1% 86.9%17 VCE 4,854 15,160 20,014 75.7% 88.5%18 EDI 1,269 14,363 15,632 91.9% 89.9%19 GLA 941 11,539 12,480 92.5% 91.1%20 GVA 843 10,655 11,498 92.7% 92.2%21 DUS 1,164 10,286 11,450 89.8% 93.3%22 STR 503 9,620 10,123 95.0% 94.3%23 HAM 617 9,187 9,804 93.7% 95.2%24 NCE 989 9,047 10,036 90.1% 96.2%25 BHX 349 6,712 7,061 95.1% 96.9%26 STN 67 5,365 5,432 98.8% 97.4%27 CGN 370 5,225 5,595 93.4% 98.0%28 BFS 213 5,062 5,275 96.0% 98.5%29 BRS 199 4,793 4,992 96.0% 99.0%30 NAP 206 3,202 3,408 94.0% 99.3%31 PSA 437 2,774 3,211 86.4% 99.6%32 BLQ 247 2,366 2,613 90.5% 99.8%33 PMO 282 1,391 1,673 83.1% 100.0%34 ORY 378 95 473 20.1% 100.0%35 LTN - - - N/A 100.0%

116

Table 4-7 shows the DB1B connecting and terminating passengers from the United States to the

35 European airports in the domain of analysis.

Table 4-8: Estimated Connecting Passenger at Top 13 European Connecting Airports in Base

Year (2007) (Source: 2007 DB1B Data; 2007 T100 Market Data).

4.9 Demonstration of Method with London, Heathrow

To demonstrate the aforementioned method, rerouting of passenger traffic from the United States

to London, Heathrow airport is shown for the year 2010 assuming the United States passenger

traffic to London, Heathrow is constrained to 2007 level. As discussed earlier in this report,

passenger traffic from the United States to Europe has been historically symmetric with traffic

from Europe to the United States.

The T100I Market data shows that 5,497,250 passengers from the United States connected or

terminated their air travel at London, Heathrow airport in 2007. Considering that the total

demand at London, Heathrow airport has reached 99.3% of the airport capacity, for this

demonstration, it is assumed that passenger traffic from the United States to London, Heathrow

airport will be held at the 5,497,250 level. The projected passengers from the United States to

London, Heathrow in 2010 is 6,609,964 resulting in an excess of 1,112,714 passengers.

European Airport

2007 DB1B Connecting

Passenger Ratio

2007 T100 Passengers from U.S.

2007 Total Connecting Passengers

(A) (B) (C = A*B)1 LHR 69.6% 5,497,250 3,825,494 2 CDG 72.9% 2,947,419 2,149,336 3 FRA 59.7% 3,265,511 1,948,110 4 LGW 85.4% 1,804,194 1,540,480 5 FCO 89.0% 742,191 660,578 6 AMS 48.9% 2,344,001 1,145,461 7 MAN 93.6% 703,639 658,760 8 MAD 74.1% 798,243 591,773 9 DUB 90.3% 716,723 647,425

10 BRU 83.9% 392,701 329,597 11 MUC 58.9% 874,898 514,909 12 ZRH 78.5% 684,262 537,418 13 BCN 89.7% 205,130 184,024

117

The candidate airports for rerouting the excess terminating passengers are London, Gatwick, 26

statute miles distant from Heathrow, London, Stansted, 41 statute miles distant, and Lodon,

Luton, 28 statute miles distant. The candidate airports for excess connecting passengers are

Charles De Gaulle, Paris France (CDG), Frankfurt, Germany (FRA) and London, Gatwick

(LGW).

From Table 4-8 the percentage of passengers from the United States connecting at London,

Heathrow is 69.6% and the percentage of terminating passengers is 30.4%. The number of

excess passengers assigned to the connecting category is 744,329 (=1,112,714 x 69.6%) and the

number assigned to the terminating category is 338,385 (=1,112,714 x 30.4%).

The number of excess terminating passengers assigned to the candidate terminating airports is

shown in Table 4-9.

Table 4-9: Assignment of the excess terminating passengers at London, Heathrow to the

candidate terminating airports.

The number of excess connecting passengers assigned to the candidate connecting airports is

shown in Table 4-10. Table 4-11 shows the resulting passenger traffic after application of the

capacity constraint and rerouting of the excess passenger traffic at all candidate airports.

Candidate Terminating

Airport

2007 T100 Market Passengers from the

United States

Percentage of 2007 T100 Market Passengers from United States at All the

Candidate Airports

Number of Excess

Terminating Passengers Assigned

London, Gatwick 1,804,194 95.1% 321,944 London, Stansted 67,965 3.6% 12,128 London, Luton 24,168 1.3% 4,313

Total 1,896,327

118

Table 4-10: Assignment of the excess connecting passengers at London, Heathrow to the

candidate connecting airports.

Table 4-11: Overview of Assignment of the excess passengers at London, Heathrow to the

candidate connecting airports and terminating airports.

4.10 Conclusions and Recommendations

Nine econometric models were developed to forecast the passenger traffic between the United

States and selected nine European countries in the period from 2008 through 2020. The total

passenger traffic from the United States to selected nine European countries is forecast to

increase from 23.2 million in 2007 to 52.7 million in 2020. The average growth rate per year is

6.5%. This does not account for the current Economic problems on both sides of the Atlantic due

to the GDP forecasts used in our analysis. The product of the United States GDP and each

Candidate Connecting Airport

2007 Connecting Passengers from the United States

(DB1B Connecting Passenger ratio * T100

Market Passengers from the United States)

Percentage of 2007 Connecting

Passengers from United States at All

the Candidate Airports

Number of Excess

Connecting Passengers Assigned

Charles De Gaulle, Paris, France 2,149,336 38.1% 295,196

Frankfurt, Germany 1,948,110 34.6% 267,559London, Gatwick 1,540,480 27.3% 211,574

Sum 5,637,926

European Airport 2010 Projected

Passengers from the

United States

Connecting Passengers

from United States

Rerouted

Terminating Passengers

from United States

Rerouted

2010 Passengers with 2007 Constraint on

United States Passengers to

London, HeathrowLondon, Heathrow 6,609,964 -774,329 -338,385 5,497,250Charles De Gaulle, Paris, France 3,498,603 295,196 3,793,799

Frankfurt, Germany 3,739,970 267,559 4,007,529London, Gatwick 2,169,386 211,574 321,944 2,702,904London, Stansted 81,722 12,128 93,850London, Lutton 29,060 4,313 33,373

Total 16,128,705 16,128,705

119

European country’s GDP was a significant variable to forecast the transatlantic passenger traffic.

The effect of September 11, 2001 on the transatlantic passenger traffic was statistically

significant, while average airfare was found not a significant variable to explain the trend of

passenger traffic.

Sixty-eight new nonstop flights between the United States airports and the European airports are

likely to be introduced by 2020 according to the demand forecast. These new nonstop flights are

suggested for two types of airport pairs: (1) those pairs which in 2007 could only be travelled

with connecting flights (2) and those pairs which in 2007 could be travelled in one flight which

includes intermediate stops. 30 new nonstops flights are suggested from 2010 on; 46 from 2015

on, and 68 from 2020 on. Analysis of 2007 airline behavior indicates that airlines are likely to

establish 12 month service for transatlantic airport pairs when the passenger demand per year is

in the neighborhood of 44,000. This translates to five or more flights per week.

The forecast passenger demands on the 182 airport pairs (with existing direct service) need to be

revised when new nonstop flights are established for 68 airport pairs. New nonstop airport pair

markets will attract some passengers traveling through the 182 airport pairs with existing direct

service in the future. The passenger demand needs to be revised to consider the competition

effect of new nonstop airport pair markets on the existing ones.

To fully analyze the airport capacity constraint on the passenger demand, both enplaning

passengers and deplaning passengers needs to be rerouted, and both the domestic passengers and

all the international passengers at the “at-capacity” airport needs to be rerouted. Since the

domain of this analysis only includes the transatlantic passenger traffic, and the data of passenger

demand for other passenger categories is limited, it is simply assumed in the above analysis that

the domestic passengers and international passengers to/from different world regions will keep

their shares in base year (2007) in the future. In the practical operations, the effect of capacity

constraint on the passenger demand of different categories (domestic/specific world region) may

vary depending on the airport/airline’s strategies.

120

5 Domestic Leg of International Passengers within the Continental U.S. (to be submitted

to Journal of Air Transport Management)

Abstract

The number of international enplanements within CONUS is predicted to grow from 74.7 million

in 2008 to 184.4 million in 2028. The average annual growth rate is expected to be 4.7%. It was

also shown that the proportion of international enplanements relative to total enplanements

within CONUS increased from 8% in 1990 to 11% in 2008. Some international passengers

generate enplanements not only at the gateway airport, but also domestic enplanements before

(departure trip) or after (arrival trip) the gateway airport. Specifically, 51% of the sampled

international and U.S. territories passengers served by U.S. carriers had at least one domestic

coupon in 2007. The objective of this chapter is to estimate the international passenger traffic

through airport-pairs within the Continental U.S. (CONUS), and to estimate the domestic

enplanements (DOI enplanements) at airports within CONUS generated by the international

passengers during the time period 1990 through 2030.

The number of DOI passengers through airport-pairs in each of the historical years (1990-2007)

is estimated based on the adjusted 100% international itineraries including pure international

itineraries plus the non-CONUS itineraries. The 100% adjusted international itineraries are

obtained based on the O&D Survey and T-100 Market data during the same time period. To

estimate the DOI enplanements at airports and the DOI passengers through airport-pair coupons

in the future years (2008-2030), the total number of DOI enplanements in each future year is

forecast first by relating it with the number of international passengers. The forecast total number

of DOI enplanements is then assigned among the commercial airports based on their shares of

DOI enplanements in base year (2007). The obtained DOI enplanements/deplanements at

airports in each future year is distributed among all the airport-pair coupons based on the matrix

of DOI passenger traffic between airport-pair coupons in base year (2007) using Fratar model.

The total number of DOI enplanements is estimated to grow from 37.3 million in 1990 to 79.4

million in 2007. The DOI enplanements were generated by 41.4 million total international and

non-CONUS passengers in 1990, and 86.3 million in 2007. The estimated DOI enplanement

generated by one international or non-CONUS passenger on average is between 0.9 and 1.1

121

during 1990 – 2007. The number of DOI enplanements is forecast to grow from 79.4 million in

2007 to 206.2 million in 2030. The average growth per year of DOI enplanements during 2008-

2030 is expected to be 4.2%. At airport level, 193 CONUS airports are estimated to have at least

10,000 DOI enplanements in 2007. Seventy-eight of them have more than 100,000 DOI

enplanements and they cover 95% of total number of DOI enplanements; ORD has the highest

number of DOI enplanements (5.4 million), followed by ATL, MIA, LAX and SFO. At the

airport-pair level, when the passenger traffic in two directions for each airport-pair is combined,

1,999 CONUS airport pairs are estimated to have more than 1,000 international or non-CONUS

passengers in 2007. LAX-ORD has the highest number of international or non-CONUS

passengers (1.1 million), followed by MCO-MIA, LAS-LAX, JFK-MIA and MIA-ORD.

122

5.1 Introduction

The objective of this chapter is to estimate the international passenger traffic through airport-

pairs within the Continental U.S. (CONUS), and to estimate the domestic enplanements (DOI

enplanements) at airports within CONUS generated by the international passengers during the

time period 1990 through 2030. International passengers are either pure international passengers

or non-CONUS passengers. Pure international passengers refer to those who travel between

airport within CONUS and airport outside of U.S. and its territories. Non-CONUS passengers

refer to those who travel between airport within CONUS and airport in Hawaii, Alaska or U.S.

territories. DOI passengers are used to represent the international passengers who generated DOI

enplanements within CONUS, i.e., the international passengers whose itineraries involve at least

one domestic leg within CONUS. The number of DOI passengers through the airport-pair is

estimated separately for two directions.

Type A – Originating international passengers

Type B – Gateway connecting international passengers

or

Type C – Non-gateway connecting international passengers

Figure 5-1: Domain of Domestic Enplanements Due to International or Non-CONUS Itineraries

(DOI Enplanements).

123

As shown in Figure 5-1, the DOI enplanements at each airport are typically generated by three

types of international passengers. It may be generated by the international passengers originating

at the airport (Type A), or by the international itineraries connecting at the airport (Type B) and

using the airport as arriving gateway airport, or the international passengers connecting at the

airport before or after the gateway airports.

In Chapter 2, the number of international (international + non-CONUS, i.e. Hawaii, Alaska and

U.S. territories) enplanements within CONUS is predicted to grow from 74.7 million in 2008 to

184.4 million in 2028. The average annual growth rate is expected to be 4.7%. It was also shown

that the proportion of international enplanements relative to total enplanements within CONUS




domestic enplanements.

Some international passengers generate enplanements not only at the gateway airport, but also

domestic enplanements before (departure trip) or after (arrival trip) the gateway airport. For

example, a symmetric international round trip starts from ROA (Roanoke, VA). It has a stopover

at ORD (Chicago, IL) and ends at PVG (Shanghai, China). One domestic enplanement is

generated at ROA by the outbound trip and at ORD by the return trip. As presented in Table 5-1,

51% of the sampled international and U.S. territories passengers served by U.S. carriers had at

least one domestic coupon in 2007, i.e., the outbound international passengers started their trips

at airports other than the gateway airports, and the inbound international passengers stopped their

trips at airports beyond the gateway airports, which results that one international passenger

generates about one domestic enplanement on average.

124

Table 5-1: Distribution of No. of Domestic Leg of International and U.S. territory Itineraries

(Data Source: 2007 DB1B Data).

No. of Domestic

Leg

Sampled Int. Passengers in

DB1B

Share of Total Sampled Int.

Passengers in DB1B

DOI Enplanements

Sampled in DB1B

Share of Total DOI Enplanements

Sampled in DB1B0 2,612,593 49.3% - 0.0% 1 485,619 9.2% 485,619 9.1% 2 1,891,624 35.7% 3,783,248 70.7% 3 178,587 3.4% 535,761 10.0% 4 106,187 2.0% 424,748 7.9% 5 14,471 0.3% 72,355 1.4% 6 5,319 0.1% 31,914 0.6% 7 1,381 0.0% 9,667 0.2% 8 499 0.0% 3,992 0.1% 9 209 0.0% 1,881 0.0%

10 67 0.0% 670 0.0% 11 28 0.0% 308 0.0% 12 12 0.0% 144 0.0% 13 3 0.0% 39 0.0% 14 2 0.0% 28 0.0% 19 1 0.0% 19 0.0%

Total 5,296,602 100.0% 5,350,393 100.0%

At airport level, as shown in Figure 5-2, the ten airports with most sample DOI enplanements

from DB1B data in 2007 is ORD, ATL, LAX, MIA, IAH, DFW, SFO, EWR, DTW and JFK in

decreasing order. The share of DOI enplanements over total sampled domestic enplanements at

seven of them was greater than 15% in 2007, and the share at MIA reached as high as 41%. DOI

enplanement was sampled at 394 airports in 2007. 91 (23%) of them had more than 5,000 sample

DOI enplanements. They covered 96.1% of total sample DOI enplanements. 20 (5%) of them

had more than 100,000 sample DOI enplanements. They covered together covers 47% of total

sampled DOI enplanements.

125


15 Airports with Most DOI Enplanements in 2007 (Data Source: 2007 DB1B Data).

At airport-pair level, as shown in Figure 5-3, the share of sample nondirectional DOI passenger

traffic over total sample DOI passenger through nine of thirty airport-pairs with most sample

nondirectional DOI passenger traffic in 2007 was greater than 40%. This share for MCO-MIA

reached 69%. The airport-pairs are in decreasing order of sample nondirectional DOI passenger

traffic from the bottom to the top. LAX-SFO had the most nondirectional sample DOI passenger

flow, followed by MCO-MIA, MIA-ORD, LAS-LAX, LAX-ORD, DFW-MIA, JFK-MIA,

DFW-LAX and JFK-LAX. DOI passenger was sampled through 6,895 airport pairs in 2007.

1,038 (15%) of them had more than 1,000 nondirectional sample DOI passengers. They covered

93.6% of total sample DOI passengers in DB1B in 2007. 161 (2.3%) of them had more than

10,000 DOI passengers and they covered 49.2% of total sampled DOI enplanements.

The publication about the estimation of either DOI enplanements at airport level or the DOI

passengers at airport-pair level is rare. Boyd group concludes that one international passenger

generated 1.5 domestic enplanements on average. The number of DOI enplanements is predicted

to grow from 114 million to 118 million in the next five year. They also indicate that 25.5% of

all domestic enplanements are generated by the international passengers. The share is predicted

to grow closer to 30% in the next five years. The methodology Boyd group used to make the

above conclusions is not publicly available.

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

ORD ATL LAX MIA DFW IAH SFO EWR JFK DTW SEA DEN IAD MSP

Share of Sam

pled

DOI

Enplan

emen

ts over A

ll Sampled

Dom

estic En

plan

emen

ts in

2007

15 Airports with Most Sampled DOI Enplanements in 2007

126


30 Airport-Pairs with Most DOI Passengers in 2007 (Data Source: 2007 DB1B Data).

5.2 Data sources

Data for this analysis includes T-100 data and the Airline Origin and Destination (O&D) Survey

data. Both data come from the Bureau of Transportation Statistics (BTS).

5.2.1 T-100 data

T-100 data is also known as “Form 41 Traffic”, “Form 41 Schedule T-100” and “Airline Market

and Segment Data”. It is collected by BTS from the Form 41, Schedule T-100 U.S. Air Carrier

Traffic and Capacity Data by Nonstop Segment and On-flight Market, and Form 41, Schedule T-

100 Foreign Air Carrier Traffic and Capacity Data by Nonstop Segment and On-flight Market.

0% 10% 20% 30% 40% 50% 60% 70% 80%

LAX‐SFOMCO‐MIAMIA‐ORDLAS‐LAXLAX‐ORDDFW‐MIAJFK‐MIALAX‐MIADFW‐LAXJFK‐LAX

EWR‐IAHORD‐SFODCA‐MIABOS‐MIAATL‐MIA

DFW‐ORDJFK‐SFOLAS‐SFOLAX‐SEAATL‐LAXATL‐MCOLGA‐MIALAX‐SANBOS‐JFK

MSP‐ORDDEN‐LAXSEA‐SFOIAH‐LAXDEN‐SFODTW‐MSP

Share of Sampled Int'l and Non‐CONUS Passenger Flow over All Sampled Passenger Flow in DB1B Data in 2007

30 Airpo

rt Pairs with Most S

ampled

Int'l & Non

‐CONUS in 2007

127

T100 data is divided into four databases: Domestic Segment, Domestic Market, International

Segment, and International Market. Domestic data covers the operations between airports located

within the U.S. and its territories. International data covers the operations between the U.S. and

foreign countries.

T100 Market data provides information includes the number of passengers, freight and mail

related to direct flight airport-pair market. Direct flight may refer to any non-stop flight or flight

that has one or multiple stopovers before reaches its destination. The difference between a direct

flight with multiple stopovers and a connecting flight is whether the passengers need to deplane

and enplane at the stopover airport. The passengers who take the connecting flight must deplane

one flight and enplane another flight at the stopover airport(s). Therefore, at least two flight

numbers are involved with the connecting flights. While the passengers on the direct flight don’t

need to deplane and enplane at the stopover airport(s), and the flight number of direct flight does

not change even though it may stop at some intermediate airport(s) to drop off and pick up

passengers.

T100 Segment data provides traffic data including the number of departures scheduled,

departures performed, passengers transported by class service, freight transported, and mail

transported), capacity data including the number of available seats departed by service class and

cargo payload capacity, and performance data including ramp-to-ramp elapsed time and airborne

elapsed time). The database provides traffic, capacity and operation information related to non-

stop flights by air carrier, airport-pair flight segment (which is also known as a flight stage or a

flight leg), and aircraft type.

According to Accounting and Reporting Directives No. 261 issued by the Office of Airline

Information of the Bureau of Transportation Statistics

(http://www.bts.gov/programs/airline_information/accounting_and_reporting_directives/number

_261.html), U.S. large certificated air carriers, small certificated air carriers, commuter air

carriers, and foreign carriers operating routes involving a U.S. airport are all required to report

Form 41 since October 2002. For both U.S. and foreign carriers, the data for joint service

operations is reported by the operating carrier, which is the air carrier in operational control of

the aircraft, i.e., the air carrier that uses its flight crew to perform the flight operation. Before

128

October 2002, U.S. small certificated air carriers and commuter air carriers were not required to

file the report. Foreign carriers didn’t need to report their operations conducted with small

aircraft (small aircraft are defined as aircraft with 60 seats or less, or 18,000 pounds or less of

payload capacity). More details about the requirement for the airlines to report Form 41 data can

be found in Title 14 of the Code of Federal Regulations (14 CFR) Part 217 and Part 241.

T100 International Market data for 1990 to 2007 is used for our empirical analysis.

5.2.2 Airline O&D Survey (DB1A/DB1B)

The Airline O&D Survey data is used to obtain the sampled air passenger itineraries. DB1B data

is collected quarterly by the U.S. Department of Transportation (DOT). It is a 10% random

sample of airline tickets from reporting carriers. Every airline ticket sold is identified by a ticket

number. It collects sample itineraries from participating carriers, i.e., U.S. large certificated air

carriers conducting scheduled passenger services (except helicopter carriers). The participating

carriers are required to report the ticket they issued or where they operated at least one flight

coupon in the ticket’s itinerary if the ticket serial number ends in “0”. Group tickets are reported

differently according to the number of the passengers on the ticket. Group tickets with ten or less

passengers on the ticket are included on the basis of 10% sample. While group tickets with more

than ten passengers on each ticket are included on the basis of a 100 percent census, regardless of

the serial number.

O&D Survey data does not provide passengers’ itineraries by non-stop flight segment. Instead, it

provides the passengers’ itineraries by coupon, showing where the passengers need to enplane to

travel from their origin and destination. A coupon may denote a nonstop flight or a single plane

through flight. A single plane through flight is a flight that stops at an intermediate airport

without changing aircraft and flight number. Besides the origin and destination airports of the

ticket’s itinerary, the data also provides the intermediate airports where the passengers need to

deplane/enplane. The other information O&D Survey data provides is the number of passengers

on the ticket, airfare (in US dollars), airfare class, operating carrier (operating carrier is an air

carrier in operational control of the aircraft, i.e., the air carrier that uses its flight crew to perform

the flight operation), ticket carrier (ticket carrier is an air carrier that issued a flight reservation or

ticket under a code share agreement regardless of their direct engagement in operation). The

129

airfare is for the whole ticket. i.e., for one way trip, the airfare is for the whole itinerary. For the

round-trip, the airfare is for the whole trip. The operating carrier and ticket carrier are for each

flight coupon in a ticket.

O&D Survey data before 1998 is also known as DB1A. DB1B data with new fare code and

record layout replaced the original DB1A data since 1998. According to Accounting and

Reporting Directives No. 224 issued by the Office of Airline Information of the Bureau of

Transportation Statistics

(http://www.bts.gov/programs/airline_information/accounting_and_reporting_directives/number

_224.html), the participating U.S. air carriers are required to report both ticket carrier and

operating carrier for each flight coupon from January, 1998. The participants were also required

to report the data with new fare code and record layout.

In case of code-share ticket, the first operating carrier is usually responsible to file the survey

data if it is a participating carrier in the O&D Survey. Therefore, although U.S. small certificated

air carriers, commuter air carriers, foreign air carriers don’t directly participate in the O&D

Survey, the data includes tickets issued by these carriers, since passengers who take flights code-

shared between these carriers and the participating carrier may be sampled by the participating

carrier.

It is important to understand how the passengers’ itineraries are reported to DOT in T-100

Segment, T-100 Market and DB1B data. For example, a hypothetical passenger flies a one-way

trip from ROA (Roanoke, VA) to LAX (Los Angeles, CA) with a stopover at ORD (Chicago,

IL). In T-100 Segment data, the trip is reported to include. However, how the trip is reported in

T-100 Market and DB1B depends on if the passenger needs to deplane and enplane at ORD,

besides enplaning at ROA and deplaning at LAX.

If the passenger needs to deplane and enplane at ORD (connecting flight),

T-100 Segment data: two non-stop segments: ROA-ORD, ORD-LAX

T-100 Market: two local markets: ROA-ORD, ORD-LAX

DB1B data: two coupons: ROA-ORD, ORD-LAX

Otherwise (single-plane through flight),

130

T-100 Segment data: two non-stop segments: ROA-ORD, ORD-LAX

T-100 Market: 1 market: ROA-LAX

DB1B data: 1 coupon: ROA-LAX

Though the trip is reported differently for the above two cases, the coupon information in DB1B

data is more consistent with the market information in the T100 Market data for the same airport-

pair.

In summary, T-100 Segment data provides information related to flight segment. A segment

refers to nonstop flight from an origin airport to a destination airport. DB1B data does not

provide passengers’ itineraries by non-stop segment. Instead, it provides the passengers’

itineraries by coupon, showing where the passengers need to enplane to travel from their origin

and destination. T100 Market data provide information related to direct flight that is consistent

with DB1B information related to coupon.

5.3 Methodology



itineraries plus the non-CONUS itineraries. The 100% adjusted international itineraries are

obtained based on the O&D Survey and T-100 Market data during the same time period. Since

we don’t have O&D Survey for forecast years (2008-2040), the number of DOI passengers

through airport-pairs during this time period is estimated differently from that of the historical

years. To estimate the DOI enplanements at airports and the DOI passengers through airport-pair

coupons in the future years (2008-2030), the total number of DOI enplanements in each future

year is forecast first by relating it with the number of international passengers. The forecast total

number of DOI enplanements is then assigned among the commercial airports based on their

shares of DOI enplanements in base year (2007). The obtained DOI enplanements/deplanements

at airports in each future year is distributed among all the airport-pair coupons based on the

matrix of DOI passenger traffic between airport-pair coupons in base year (2007) using Fratar

model.

131

5.3.1 Estimation of DOI Passengers through Airport-Pairs during 1990-2007



itineraries plus the non-CONUS itineraries. Though 100% international itineraries are not

available directly during 1990-2007, 10% random airline tickets are available from O&D Survey

for each of the historical years. The 100% level of international itineraries in each year are

obtained by adjusting the number of passengers for each DB1B sampled international itinerary

using the scale factors defined by gateway airport. It is assumed that the number of DOI

passengers who buy separate tickets for their domestic trips within U.S. is small. Because of

lower combined airfare, some DOI passengers buy separate tickets for segments of their whole

trips. Since we cannot tell the domestic trips are actually part of international or non-CONUS

trips from DB1B data that is ticket based, the DOI enplanements generated by those DOI

passengers who buy separate tickets for part of their whole trips are not captured by our

estimation though they generate domestic enplanements.

First of all, all the sample itineraries in DB1B data are divided into pure domestic and

international itineraries based on the world area codes (WACs) of the airports included in the

itinerary. Pure domestic itineraries are excluded in the analysis. Based on the world area code, an

airport is identified as CONUS (WAC > 5 and ≤ 99), international (WAC > 99) or non-CONUS

(WAC ≤ 5). An itinerary is identified as pure domestic itinerary if all the airports included in the

itinerary are pure domestic. In 2007, 10.9 million (70%) of total 15.6 million tickets sampled in

DB1B data represent pure domestic itineraries. The rest of the itineraries involving at least one

pure international airport or non-CONUS airport are referred as international itineraries.

DB1B data including itineraries involving only domestic (pure domestic or non-CONUS) airport

is available publicly at BTS website in three tables: DB1B Coupon, DB1B Market and DB1B

Ticket to provide coupon-specific information, directional market characteristics and summary

characteristics of each itinerary, respectively. We convert DB1B data into Coupon, Market and

Ticket using the same table profiles. The Matlab function used to divide and convert the raw

DB1B data is attached in Appendix D.1. The input of the function is the quarterly raw DB1B

data, the output is six databases including Pure Domestic Coupon, Pure Domestic Market, Pure

Domestic Ticket, International & non-CONUS Coupon, International & non-CONUS Market,

132

and International & non-CONUS Ticket. International & non-CONUS Coupon is used for the

rest of our analysis because it includes the most detailed information for all sample itineraries

involving at least one international or non-CONUS airport.

For each of the international itineraries, the gateway coupon is identified as the coupon whose

origin is a CONUS airport and destination is outside of CONUS, i.e., an international or non-

CONUS airport. An itinerary is identified as pure international itinerary if the airport outside of

CONUS is international or non-CONUS itinerary if the airport out of CONUS is non-CONUS.

When an itinerary has multiple gateway coupons, the airport outside of CONUS of the first

gateway coupon is used to identify the itinerary type. Some sample international itineraries don’t

include any gateway coupon. For example, the one-way inbound itineraries or the itineraries

involving trips between non-CONUS and international airport. These itineraries are excluded in

our analysis.

The DOI enplanements generated by pure international passengers and non-CONUS passengers

are estimated separately.

5.3.1.1 Pure International Itineraries

Pure International itineraries are divided into U.S. carrier and foreign carrier based on the

operating carrier of the gateway coupon. Considering that domestic leg information of the

foreign carrier itineraries sampled in DB1B data are biased, only U.S. carrier sample

international itineraries are used in the analysis. The domestic legs of foreign carrier international

itineraries are assumed to follow the pattern of U.S. carrier international itineraries.

As introduced in “Airline O&D Survey” Section, although the foreign air carriers do not directly

participate in the O&D Survey, DB1B includes some of their data, since passengers who take

flights code-shared between U.S. participating carrier and foreign carrier may be sampled by the

U.S. carrier. For example, Air Canada operates the first coupon of an itinerary, and interlines and

code-shares with United Airlines. The itinerary may be reported by United Airlines.

The foreign carrier sample itineraries are biased because none of the foreign carrier itineraries is

required to be reported in the O&D Survey when no U.S. participating carrier operates any

coupon of the itineraries. For example, the single coupon itinerary from ATL(Atlantic, GA) to

133

LHR (London, UK) both operated and ticketed by a foreign carrier, or the same single coupon

itinerary operated by a foreign carrier under a U.S. carrier (ticketed carrier), both are missing in

the DB1B data. But in both cases, the enplanements are captured as foreign carriers’ at ATL in

T100 data. Because we use the enplanements obtained from T100 data as benchmark to adjust

the sampled itineraries to 100% level, to make sure that the samples are random, the sampled

foreign carrier itineraries in DB1B are excluded in the analysis.

Total 2,950,933 sample U.S. carrier pure international itineraries using significant gateway

airport are adjusted to obtain the 100% pure international itineraries. Gateway airports with more

than 10,000 total pure international enplanements and more than 1,000 DB1B sample pure

international enplanements, both operated by U.S. carriers, are considered significant each year.

Most gateway airports identified from T100 data have little contribution to the total international

enplanements. For example, total 156 gateway airports for international trips were identified

from T100 data in 2007, but 117 of them had less than 10,000 enplanements and these 117

airports cover only 0.2% of total international enplanements. Therefore, the gateway airports

with less than 10,000 international enplanements are excluded in the analysis. On the other side,

some of the gateway airports with more than 10,000 enplanements have too few sample

itineraries in DB1B data. For example, in 2007, 13,366 international enplanements were

recorded at LKE (Lake Union Seaplane Base, WA) from T100 data, but none international

enplanement at LKE was sampled in DB1B. To make sure that only gateway airports with

reliable sampled itineraries are used. The gateway airports with less than 1,000 sampled DB1B

enplanements are also excluded in the analysis.

Two scale factors are estimated for each significant gateway airport for pure international

itineraries. _ is used to adjust DB1B sampled U.S. carriers international itineraries

to 100% U.S. carriers international itineraries, and _ is used further to adjust the

100% U.S. carriers international itineraries to 100% total (U.S. carriers + foreign carriers)

international itineraries. To obtain the DB1B sampled U.S. carriers international enplanements at

each of significant gateway airports, the number of passengers for the sampled international U.S.

carriers itineraries is aggregated by the corresponding gateway airport.

_100%_ _ _

1 _ _ _

134

_100%_ _ _

100%_ _ _

where

_ – Scale factor used to adjust DB1B sampled U.S. carriers international

itineraries to 100% U.S. carriers international itineraries using significant gateway airport i

_ – Scale factor used to adjust 100% U.S. carriers international itineraries to

100% total (U.S. carriers + foreign carriers) international itineraries using significant

gateway airport i

100%_ _ _ – 100% U.S. carriers international enplanements at

significant gateway airport i from T100 International Market data

1 _ _ _ – Sampled U.S. carriers international enplanements at

significant gateway airport i from the DB1B data

100%_ _ _ – 100% international enplanements at significant

gateway airport i from T100 International Market data

In 2007, thirty-nine significant gateway airports were identified with more than 10,000 pure

international enplanements and more than 1,000 DB1B sample pure international enplanements.

The scale factors for these gateway airports are shown in Figure 5-4 and Figure 5-5. As

presented, the estimated _ for most of the gateway airports for pure international

itineraries are close to ten, which are expected since DB1B data randomly samples 10% of

airline tickets. While the scale factors for PBI (West Palm Beach, FL), CLE (Cleveland, OH) and

FLL (Fort Lauderdale, FL) are higher than 13. Especially, the scale factor for PBI is as high as

29. Their high scale factors are resulted from the high share of pure international passengers

operated by commuter carriers that are required to file T100 data but not required to participate

in the O&D Survey. Specifically, 99.8% of pure international passengers were operated by

Gulfstream International Airlines at PBI, 13% by Gulfstream International Airlines at FLL, and

18% by Chautauqua Airlines at CLE. Both Gulfstream International Airlines and Chautauqua

Airlines are required to file T100 data but not required to participate in the O&D Survey.

135

Figure 5-4: _ for Thirty-Nine Gateway Airports for Pure International Itineraries.

0 5 10 15 20 25 30 35

MIAJFKATLEWRORDIAHDFWLAXSFODTWPHLIADMSPCLTFLLSEADENPHXBOSCVGSLCLGA

MEMLASPDXCLE

MCOBWISANRDUSTL

MDWPITDCAPBIBDLMCIMKEIND

Scale_Factor1 in 2007

39 Gatew

ay Airpo

rts (Int'l)

136

Figure 5-5: _ for Thirty-Nine Gateway Airports for Pure International Itineraries in

2007.

5.3.1.2 Non-CONUS Itineraries

740,346 sample non-CONUS itineraries using significant gateway airport are adjusted to obtain

the 100% non-CONUS itineraries. Unlike pure international itineraries, all the non-CONUS

itineraries are operated by U.S. carriers. Therefore, all the sampled non-CONUS itineraries are

considered random. Additionally, the itineraries adjusted using _ is the 100% of

total Non-CONUS itineraries, so no _ is needed. All the sample non-CONUS

itineraries using significant gateway airport are adjusted to obtain the 100% of non-CONUS

itineraries. Like the analysis of international itineraries, only gateway airports with more than

0 2 4 6 8 10 12 14 16

MIAJFKATLEWRORDIAHDFWLAXSFODTWPHLIADMSPCLTFLLSEADENPHXBOSCVGSLCLGA

MEMLASPDXCLE

MCOBWISANRDUSTL

MDWPITDCAPBIBDLMCIMKEIND

Scale_Factor2 in 2007

39 Gatew

ay Airpo

rts (Int'l)

137

10,000 total non-CONUS enplanements and more than 1,000 DB1B sample non-CONUS

enplanements are considered significant each year.

In 2007, thirty-three significant gateway airports were identified. These airports covered 99.7%

of all the sampled non-CONUS passengers from DB1B and also 99.7% of all the non-CONUS

enplanements from T100. As presented in Figure 5-6, the _ for thirty-three

gateway airports for US territory itineraries are ranged between 8.9 and 11.2. It shows that the

sample share of US territory itineraries by DB1B data is very close to 10%.

Figure 5-6: _ for Thirty-Three Gateway Airports for the U.S. Territory Travel.

5.4 Forecast of DOI during 2008-2030

To estimate the DOI enplanements at airports and the DOI passengers through airport-pair

coupons in the future years (2008-2030), the total number of DOI enplanements in each future

year is forecast first by relating it with the number of international passengers. The forecast total

0 2 4 6 8 10 12

LAXSEASFOJFKMIAMCOORDATLEWRPHXDFWMSPOAKFLLLASIAHPDXPHLSLCBOSSANDENCLTSMFIADSNATPASJCBDLBWICVGONTDTW

Scale Factor in 2007

33 Gatew

ay Airpo

rts (US Territories)

138

number of DOI enplanements is then assigned among the commercial airports based on their

shares of DOI enplanements in base year (2007). The obtained DOI enplanements/deplanements

at airports in each future year is distributed among all the airport-pair coupons based on the

matrix of DOI passenger traffic between airport-pair coupons in base year (2007) using Fratar

model.

Figure 5-7: Thirty-Three Gateway Airports for the U.S. Territory Travel.

As evidenced in Figure 5-7, the estimated DOI enplanements during 1990 and 2007 are highly

correlated with the number of international passengers during the same period. It was found that

one international passenger is responsible for 1.1 DOI enplanements on average, which is lower

than what Boyd group predicted (1.5 DOI enplanements per international passenger) for the next

five year. The forecast number of international passengers obtained from Chapter 2 and its

relationship with the total number of DOI enplanements are applied to estimate the total number

of DOI enplanements in the NAS in the future years.

The growth of total DOI enplanements is applied equally to all the commercial airports in the

NAS. It is assumed that the share of DOI enplanements/deplanements at each airport over total

DOI enplanements is constant for the future years. This assumption is considered reasonable

based on the relatively stable share at the airports with most DOI enplanements in recent years.

The share of DOI enplanements at top 30 airports over total DOI enplanements within the

CONUS is presented for the time period of 1998-2007 in Figure 5-8. These 30 airports cover

81% of total DOI enplanements in 2007. Though the share of few airports experienced decrease

or increase, the share of most of the airports are stable after year 2005.

y = 0.8917x + 8E+06R² = 0.9017

25

35

45

55

65

75

85

95

40 50 60 70 80 90

Dom

estic En

plan

emen

ts due

to

Int'l and

U.S. Territory Passeng

ers

during

1990 ‐2007 (M

illions)

International & U.S. Territory Enplanements during 1990 ‐ 2007 (Millions)

139

Figure 5-8: Top 30 Airports’ Share of Total DOI Enplanements within the CONUS in 2007.

The obtained DOI enplanements/deplanements at airports in each future year is distributed

among all the airport-pair coupons based on the matrix of DOI passenger traffic between airport-

pair coupons in base year (2007) using Fratar model. The details of Fratar model can be found in

Chapter 4.

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

1998 1999 2000 2001 2002 2004 2005 2006 2007

Airpo

rts (Top 15)' Sh

are

of Total DOI Enp

lane

men

ts in

CONUS

Year

ORD

ATL

MIA

LAX

SFO

DFW

IAH

JFK

EWR

SEA

DEN

DTW

IAD

LAS

PHL

0%

1%

1%

2%

2%

3%

3%

1998 1999 2000 2001 2002 2004 2005 2006 2007

Airpo

rts (16

‐30t

h )' Sha

re

of Total DOI Enp

lane

men

ts in

CONUS

Year

MCO

MSP

BOS

PHX

CLT

LGA

SLC

SAN

DCA

PDX

TPA

CVG

FLL

RDU

CLE

140

5.5 Results

5.5.1 The Total Number of DOI Enplanements

The total number of historical (1990-2007) DOI enplanements is estimated by aggregating the

corresponding adjusted 100% international and non-CONUS itineraries for each year. The

forecast (2008-2030) DOI enplanements is obtained by the forecast number of total international

passengers, and the close relationship between the number of DOI enplanements and the number

of international passengers. As presented in Figure 5-9, the total number of DOI enplanements is

estimated to grow from 37.3 million in 1990 to 79.4 million in 2007. The DOI enplanements

were generated by 41.4 million total international and non-CONUS passengers in 1990, and 86.3

million in 2007. The estimated DOI enplanement generated by one international passenger on

average is between 0.9 and 1.1 during 1990 – 2007. The number of DOI enplanements is forecast

to grow from 79.4 million in 2007 to 206.2 million in 2030. The average growth per year of DOI

enplanements during 2008-2030 is expected to be 4.2%.

Figure 5-9: Historical (1990-2007) and Forecast (2008-2030) DOI Enplanements within the

CONUS.

5.5.2 The Number of DOI Enplanements at Airports

In 2007, 193 CONUS airports are estimated to have at least 10,000 DOI enplanements in 2007.

Seventy-eight of them have more than 100,000 DOI enplanements and they cover 95% of total

number of DOI enplanements; Twenty-three airports estimated to have more than one million

DOI enplanements are presented in Figure 5-10. These twenty-three airports together cover 75%

‐

50

100

150

200

250

1990 1995 2000 2005 2010 2015 2020 2025 2030

Millions

Year

International Enplanements

Enplanements due to International & U.S. Territory Travel

Base Year

141

of total number of DOI enplanements. The airports are in the decreasing order of DOI

enplanements from the bottom to the top. ORD has the highest number of DOI enplanements

(5.4 million), followed by ATL with 4.8 million, MIA with 4.7 million, LAX with 4.3 million

and SFO with 3.7 million.

Figure 5-10: Twenty-Three Airports with More Than One Million DOI Enplanements within the

CONUS in 2007.

The historical (1990-2007) and the forecast (2008-2030) trend of DOI enplanements at the top

five airports are presented in Figure 5-11.

0 1 2 3 4 5 6

ORDATLMIALAXSFODFWIAHJFK

EWRSEADENDTWIADLASPHLMCOMSPBOSPHXCLTLGASLCSAN

DOI Enplanements in 2007 (Millions)

23 CONUS Airpo

rts w

ith more than

one

million

DOI Enp

lane

men

ts in

2007

142

Figure 5-11: Historical (1990-2007) and Forecast (2008-2030) DOI Enplanements at the Top 5

Airports.

5.5.3 The Number of DOI Passengers through Airport-Pairs

When the passenger traffic through each airport-pair in two directions is combined, 1,999 airport

pairs are estimated to have more than 1,000 international or non-CONUS passengers in 2007.

1,088 airport-pairs have more than 10,000 passengers and covered 95% of total DOI

enplanements. 199 of them have more than 100,000 and they cover 57% of total DOI

enplanements. Thirty CONUS airport-pairs with most international or non-CONUS passengers

are presented in Figure 5-12. The passenger traffic in both directions is displayed for each

airport-pair. These thirty airports together cover 81% of total DOI enplanements. The airport-

pairs are in the decreasing order of the sum of passenger traffic in two directions from the bottom

to the top. LAX-ORD has the highest number of international and non-CONUS passengers (1.1

million), followed by coupons MCO-MIA with 0.97 million, LAS-LAX with 0.88 million, JFK-

MIA with 0.79 million and MIA-ORD with 0.76 million. It is noted that international or non-

CONUS passenger traffic are not balanced in both directions for some of the airports. This may

be resulted from the one-way tickets or some asymmetric itineraries of round trip tickets. The

sampled international or non-CONUS tickets with odd number of domestic leg is 13% of the

total DB1B sampled international or non-CONUS tickets in 2007.

‐

2

4

6

8

10

12

14

16

1990 1995 2000 2005 2010 2015 2020 2025 2030

Millions

Year

ORD

ATL

MIA

LAX

SFO

Base Year

143

Figure 5-12: Thirty CONUS Flight Coupons with Most International & Non-CONUS Passenger

Traffic in 2007.

The estimated historical (1990-2007) and the forecast (2008-2030) trend of the sum of

international passenger traffics in two directions are presented for the top five airport-pairs in

Figure 5-13.

0.0 0.2 0.4 0.6 0.8 1.0 1.2

LAX‐SFOMCO‐MIALAS‐LAXJFK‐MIA

MIA‐ORDJFK‐LAX

DFW‐MIALAX‐MIALAX‐SEAEWR‐IAHLAX‐ORDBOS‐MIADCA‐MIAORD‐SFOJFK‐SFOLGA‐MIALAS‐SFOBOS‐JFK

ATL‐MCOATL‐MIA

DFW‐ORDJFK‐LGALAX‐SANDEN‐SFODFW‐LAXSEA‐SFOIAD‐SFODEN‐LAXEWR‐MIAMIA‐TPA

DOI PassengerTaffic in 2007 (Millions)

30 Cou

pons with Most D

OI Passeng

er Traffic in

2007

Sum of two directions

Direction reversed

Direction as specified

144

Figure 5-13: Historical (1990-2007) and Forecast (2008-2030) of Sum of International or Non-

CONUS Passenger Traffic in Two Directions for Top 5 Airport-Pairs.

5.6 Conclusions

The total number of DOI enplanements is estimated to grow from 37.3 million in 1990 to 79.4

million in 2007. The DOI enplanements were generated by 41.4 million total international and

non-CONUS passengers in 1990, and 86.3 million in 2007. The estimated DOI enplanement

generated by one international or non-CONUS passenger on average is between 0.9 and 1.1

during 1990 – 2007. The number of DOI enplanements is forecast to grow from 79.4 million in

2007 to 206.2 million in 2030. The average growth per year of DOI enplanements during 2008-

2030 is expected to be 4.2%. At airport level, 193 CONUS airports are estimated to have at least

10,000 DOI enplanements in 2007. Seventy-eight of them have more than 100,000 DOI

enplanements and they cover 95% of total number of DOI enplanements; ORD has the highest

number of DOI enplanements (5.4 million), followed by ATL, MIA, LAX and SFO. At the

airport-pair level, when the passenger traffic in two directions for each airport-pair is combined,

1,999 CONUS airport pairs are estimated to have more than 1,000 international or non-CONUS

passengers in 2007. LAX-ORD has the highest number of international or non-CONUS

passengers (1.1 million), followed by MCO-MIA, LAS-LAX, JFK-MIA and MIA-ORD. It is

noted that international or non-CONUS passenger traffic are not balanced in both directions for

some of the airports. This may be resulted from the one-way tickets or some asymmetric

itineraries of round trip tickets.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

1990 1995 2000 2005 2010 2015 2020 2025 2030

Millions

Year

LAX‐SFO

MCO‐MIA

LAS‐LAX

JFK‐MIA

MIA‐ORD

Base Year

145

6 Development of Secondary Airports in Multi-Airport Systems The Impact of the EU-

US Open Skies Agreement on Commercial Airline Passenger Traffic over the North

Atlantic in the NAS (to be submitted to Journal of Air Transport Management)

Abstract




objective of this analysis is to understand the conditions under which the secondary airports

develop in a metropolitan area. The markets and connectivity supplied by the secondary airports

are also identified and compared with that of the primary airports. This analysis is needed to

understand and improve the dynamics of the network evolver developed for the Transportation

System Analysis Model (TSAM).

For this study, fifteen multi-airport systems including 19 Operational Evolution Plan (OEP)

airports and 25 active secondary airports are identified in the NAS. 59 inactive secondary

airports are located around 32 OEP airports. The traffic is more volatile at the inactive secondary

airports compared with that at the active secondary airports. A threshold value of domestic

Originating and Destination (O&D) passengers (or total enplanements) in the metropolitan area

that produce successful secondary airports was loosely correlated with that suggested in the

literature. The catchment population share of the secondary airports did not show an increasing

trend from 1990 to 2000.

Diverse trends of traffic distribution among airports in the same metropolitan area are observed.

Among 15 MASs considered in the analysis, the enplanements share of secondary airports

increased in nine of them during the period 1990-2008. The secondary airports’ enplanements

share decreased in the following MASs: Houston, Cleveland, Tampa, Orlando and Philadelphia.

In the MAS of Miami and Tampa, the trend of enplanements share at the two secondary airports

is different. The enplanements share of one secondary airport increased while the share at the

other secondary airport in the same MAS decreased or remained flat. The enplanements share of

active secondary airports at seven MASs was higher than 18% in 2008.

146

As expected, the number of markets served at the secondary airports is less than that at the

primary airport in the same metropolitan area. The secondary airports tend to serve the short- or

medium-haul high density markets in large metropolitan areas. The secondary airport seats share

for the short- or medium-haul markets is higher than the average seat level for the primary

airport. It is lower than the average level for the long-haul markets, especially for the

transcontinental markets. Airlines provide service at multiple airports in the same metropolitan

area. Full-service Carriers (FSCs) concentrate their service mainly on the primary airport (i.e.,

hubs), while Low-Cost Carriers (LCCs) often concentrate their service on the secondary airports,

but they also select primary airports as their focus city. The distribution of the number of seats

offered by the low-cost carriers in one metropolitan area is more spread than that of FSC carriers.

Most of the secondary airports are currently dominated by the LCCs. The share of seats supplied

by the LCCs at the secondary airports increased in the period 1990-2008. The growth of traffic at

some of the secondary airports is not only directly due to the fast growth of LCCs, but also by

the traffic stimulation provided by the low-cost carrier and the response of full-service carriers.

This phenomenon is known as the “Southwest effect”.

Because FSCs rely heavily on hub-and-spoke route structures, and the primary airport in a MAS

usually functions as one or more FSCs’ hub, FSCs concentrate their service mainly on the

primary airport at all the MASs analyzed. They tend to provide short- or medium-haul flights

from the secondary airport(s) to their hub(s) or secondary hub(s) when they supply service at the

secondary airport.

The average seating capacity per aircraft at the secondary airports is higher than that of primary

airports in most of the MASs studied. The share of flights served by FSCs with turboprop and

regional jets with 37-99 seats at the secondary airports are higher than that at primary airports in

2008. Most of the secondary airports do not offer long-haul flights although all the 25 secondary

airports studied meet the runway length requirement of long-haul flights.

The secondary airports serve the domestic O&D passengers. Exceptions are MDW (Chicago

Midway), HOU (Houston/Hobby) and DAL (Dallas/Love Filed). These airports have a

significant share of connecting passengers. All other secondary airports mainly serve O&D

passengers.

147

6.1 Introduction


U.S. Secondary airports are regional airports that have been considered as a possible source of

relief to reduce airways and airport congestion at the hub airport and to efficiently accommodate

future air travel demand. Because of the fast growing travel demand, the hub-spoke airline

network causes the over demand and severe delay at the hub airports. The excessive congestion

and delays at the airport cause extra costs to both airlines and passengers. Constructing new large

airport or runways at the large airports is constrained by land use polity and limited by the strong

environmental constraints. It thus creates incentives for airlines and passengers to increase use of

underused regional airports located within reasonable proximity of larger, congested airports. For

example, SFO was one of the most heavily delayed airports in the U.S. throughout the 1990’s.

These delays contributed to service expansion and increased traffic shares at OAK and SJC.

OAK increased its share of Bay Area domestic O&D passengers from 20% in the late 1990’s up

to a peak of 33% from 2003 to 2006. SJC share gains were less pronounced (from approximate

22% up to 26% in 2002 (SH&E, 2009).

Low-cost carriers (LCCs) have expanded their service in the last decade. The share of air

passengers carried by LCCs in the National Airspace System (NAS) has increased steadily from

17% in May 2000 to 33% in May 2009. LCCs often serve their passengers using the secondary

airports in major metropolitan areas. LCCs reduce the service costs by offering no-frills service

and through, frequent flyer clubs and seat allocation. The lower operation cost compared with

full service carriers (FSCs) allows low air fares. Their point-to-point business model doesn’t

require baggage administration and direction to connecting flights. LCCs also reduce the ticket

booking costs by selling tickets on their own websites. They prefer less congested secondary

airports instead of congested airports or airports having high facility charges. Additionally, the

higher fuel efficiency and lower repair costs of LCCs is possible by using the new generation

aircraft. For example, the average fleet age of JetBlue Airways was 4.6 years. In contrast with

FSCs who use various aircraft type, LCCs usually operate single aircraft type. It provides them

advantages in terms of minimum maintenance, spare parts inventory and the pilot training cost.

To increase their share of the total market, the regional airports must be attractive for both

passengers and airlines to change their travel and service patterns respectively. The literature

148

review has showed that the airport’s ground access time and the frequency of direct flights at the

airport are two most important factors affecting the airport choice in a MAS. The demand for

low-cost or point-to-point services is the most important factor LCCs consider before deciding to

start service at an airport (Warnock-Smith and Potter, 2005). Other airport choice factors

considered include low airport charges, opportunity for quick turnarounds, simple terminals,

rapid check-in facilities, good passenger facilities and accessibility. The secondary airports offer

the following advantages compared with primary airports to LCCs and their passengers.

Excessive capacity is available and congestion is small or absent at the regional airports making

it possible for LCCs to follow tight schedules. This also reduces congestion related delay costs.

LCCs don’t rely on connections. The shorter turnaround time help LCCs keep their aircraft

airborne as much as possible. There are no problems with the availability of convenient slots

throughout the day. This allows LCCs to improve their operational efficiency by maximizing

their fleet utilization. Infrastructures such as check-in counters and handling systems at the

regional airports largely absent making it possible to design ones that are simple enough to fulfill

LCCs’ needs for quick services. Given that airport charges accounts for 12% of LCCs operating

cost (Doganis, 2001), the lower aeronautical charges at the regional airports make them more

attractive compared to primary airports. In some cases, because the local unemployment in areas

where secondary airport are located is higher, the surrounding regional authorities are willing to

offer subsidies to attract more carriers to promote the local economic development and

employment.

The objective of this analysis is to study and understand the conditions under which secondary

airport(s) develop in a metropolitan area. The markets and connectivity supplied at the secondary

airports are also identified and compared with that at the primary airports. This analysis is

needed to understand and improve the dynamics of the network evolver developed for the

Transportation Systems Analysis Model (TSAM).

149


A literature review about multi-airport systems was conducted as part of this study. The review

focused on the analysis of the dynamics of evolution of multi-airport system and the passengers’

airport choice in a multi-airport system.

6.2.1 Dynamics of Evolution of Multi-Airport System

Bolgeri et al. (2008) studied the factors that influence the development of multi-airport system.

The author proposed three main factors that affect the growth of an airport: a) technical (such as

airport capacity constraints), b) economic (such as the changing business models of the airlines

operating at the airport) and c) political/historical (such as decisions to add the airport system

capacity). The development of the world’s two largest multiple airport systems in London and

New York metropolitan areas were studied. It was concluded that the capacity constraints at

primary airports are technical effects and the entry of low-cost carries and their subsequent

growth are an economic effect driving the development of multiple airport system. Political

factors can both stimulate and constrain the multi-airport system.

Bonnefoy and Hansman (2005) also studied the factors underling the development of multi-

airport system. Five main factors were identified as driving forces in this study: a) congestion at

the primary airport, b) entries of air carriers (especially low cost carriers at the secondary

airports), c) the population distribution in the metropolitan area, d) the runway length (greater

than 5700 ft) at secondary airports, and e) the ratio of connecting passengers at the primary

airport (less than 25%). This study also suggested that new constraints are imposed considering

the impact of multi-airport system operations on NAS structure.

Considering that the traffic distribution among multiple airports serving a metropolitan area is a

result of the dynamic competition between both airlines and airports; and that the traffic at

secondary airports is volatile comparing with the traffic at primary airports, De Neufville (1995)

suggested that the development of multi-airport system are strategic, incremental and flexible.

The author mentioned that an airport was more attractive to originating passengers when it

provides convenient accessibility and high frequency of departures; and severe congestion at the

primary airport, and high traffic demand in the specific metropolitan area made the secondary

150

airports attractive to airlines. The author also mentioned that the traffic at the secondary airports

was usually of different character from traffic at primary airport.

Bonnefoy and Hansman (2005) used a System Dynamics approach to capture the main dynamics

of multi-airport system. The model included two main composite variables, i.e. the airport

attractiveness to airlines and the airport attractiveness to passengers, and four major feedback

loops, i.e. the airport growth loop, the demand stimulation loop, the airport congestion loop and

the capacity adjustment loop. This model did not attempt to validate data of a real system.

Warnock-Smith and Potter (2005) studied the ranking of factors the European low-cost carriers

considered when select the secondary airport to open new service. Based on an exploratory

survey of eight European low-cost carriers, it was found that airline ranked the demand for low-

cost service on top of other fourteen airport choice factors. Convenient slot times and lower

turnaround time ranked second; cheap aeronautical charges ranked fourth and high level of

airline competition ranked lowest. The study showed that the importance of each airport choice

factors differed among airlines of different characteristics.

6.2.2 Airport Choice Models in Multi-Airport System

Windle and Dresner (1995) modeled the passenger airport choice in the Washington/Baltimore

multi-airport system using the multinomial logit model. The business and non-business

passenger airport choice was modeled separately. They concluded that the airport access time

and flight frequencies were the two most important factors that influence passenger decisions

airport choice in the multi-airport system. It was found that shorter airport access times and

higher flight frequencies were more important for business passengers than for non-business

passengers (i.e., related to value of time). A passenger who had experience with an airport was

more likely to choose the airport again when the attributes of the airports did not change. The

airport access time was found to be less important for passengers in competitive aviation zones

while the flight frequencies was more important.

Cohas et al. (1995) modeled the passenger distribution between airports in a multi-airport system

by relating an airport’s market share with airline fares and flight frequency. The market share of

competing airports in a multi-airport system was estimated at origin-destination (O-D) level.

151

Their case studies included three major multi-airport systems: New York/New Jersey, San

Francisco Bay Area and Washington/Baltimore.

Pels et al. (1995) modeled the passenger joint airport and airline choice in the San Francisco Bay

Area multi-airport system using both multinomial logit and the nested logit model. They

concluded that the model fit of the nested logit model was better than that of the multinomial

logit model. Flight frequency and access time were found to influence passenger airline and

airport combination choices. Business and leisure passengers were modeled separately and the

difference in behavior between these two groups of passengers was found small.

Hess and Polak (2006) modeled the passenger joint airport, airline and access mode choice in the

San Francisco Bay Area multi-airport system using both multinomial logit and the nested logit

model. The passengers were segmented into six groups by their residence status (resident, non-

resident) and trip-purpose (business, holiday, visiting friends and family) and were modeled

separately. They found that the in-vehicle access time and flight frequencies were most important

two factors influencing passenger simultaneous choice of airport, airline and access mode in the

multi-airport system. The airline fares and aircraft size were only important for some of the

passenger subgroups. They also found that the multinomial logit model was found to show high

level of prediction capability, but the model fit could be improved by using a two-level nested

logit models.

Rather than modeling based on passengers’ decision structure and the characteristics of airports

in multi-airport system, Tien and Schonfeld (2007) developed an airport choice model for a

hypothetical multi-airport system assuming that the airport price charged to airlines affects the

airline flight frequencies and hence passenger airport choice. Based on this assumption, the

airport was modeled to maximize its payoff function defined to reflect both the airlines and the

passengers’ responses to its strategies of prices charged to airlines.

The previous models of airport choice in multi-airport system are summarized in Table 6-1.

152

Table 6-1: Summary of Previous Airport Choice Models in Multi-Airport System.

Literature Model Explanatory Variable Population Segment Data Hess et al. (2007)

Multinomial Logit (MNL)

Air fare, access time, flight time and airline and airport allegiance; Non-linear transforms of the explanatory variables; Treatment of continuous variations in choice behavior across respondents

‐ Business ‐ Holiday ‐ Visiting friends or

relatives (VFR)

Stated preference survey

Hess and Polak (2006)

Multinomial Logit (MNL) Nested Logit (NL)

Logarithmic transform ‐ Flight frequency ‐ Airport initial variable (flights

took in last 12 months) Linear ‐ In-vehicle access time ‐ (Fare) ‐ (Aircraft size)

Residency status ‐ Resident ‐ Non-resident Trip purpose ‐ Business ‐ Holiday ‐ Visiting friends or

relatives (VFR) Income ‐ Low (<$21,000 per

annum) ‐ Medium (<$44,000 per

annum) ‐ High (>$44,000 per

annum)

1995 Airline Passenger Survey (Bay Area MTC)

Hess and Polak (2005)

Mixed Multinomial Logit (MMNL)

Deterministic ‐ Air fare ‐ Flight frequency Random distributed ‐ Access time

Residency status ‐ Resident ‐ Non-resident Trip purpose ‐ Business ‐ Leisure Income ‐ Low (<$21,000 per

annum)


153

‐ Medium (<$44,000 per annum)

‐ High (>$44,000 per annum)

Basar and Bhat (2004)

Multinomial Logit (MNL) Probabilistic choice set multinomial logit (PCMNL)

‐ Access time ‐ Flight frequency

‐ Business 1995 Airline Passenger Survey (Bay Area MTC)

Pels et al. (2003)

Nested Logit (NL) 1. Airport 2. Access mode

Airport ‐ Air fare ‐ Flight frequency (logarithmic

transformed) Access mode ‐ Access time ‐ Access cost

‐ Business ‐ Leisure


Pels et al. (2001)

Nested Logit (NL) 1. Airport 2. Airline

Airport ‐ Access time ‐ Flight frequency (logarithmic

transformed) Airline



Windle and Dresner (1995)


‐ Access time ‐ Flight frequency ‐ Airport initial variable


1987 passenger survey (Metropolitan Washington Council of Governments & Maryland DOT)

Furuichi and Koppelman (1994)

Nested Logit (NL) 1. Depart airport 2. Destination airport

Depart airport ‐ Access time ‐ Access cost ‐ Line-haul time ‐ Line-haul cost ‐ Relative flight frequency Destination airport


1989 international air travelers behavior survey (Japanese Ministry of Transport)

154

‐ Log sum of access and line-haul service

‐ Trade value (logarithmic transformed)

Harvey (1987)


Access time (nonlinear) Flight frequency (nonlinear)

Residents ‐ Business ‐ Leisure


155

6.3 Identification of Multi-Airport System in the NAS

Figure 6-1: Number of Secondary Airports for 34 OEP Airports in the CONUS

(Source: 1990-2008 T-100 Segment).

A multi-airport system (MAS) is defined as a set of two or more significant airports that serve

commercial traffic within a metropolitan region. There is no formal definition of multi-airport

system. De Neufville and Odoni (2003) defined a multi-airport system as “the set of significant

airports that serve commercial transport in a metropolitan region, without regard to ownership or

political control of individual airports”. The set of the airports inside the MAS are classified into

primary and secondary airport. The primary airport is defined as an airport serving most of the

traffic in a multi-airport system or an airport having at least 50% of traffic that the most served

airport has. Usually, the secondary airport is the underused airport that complements the primary

airport in a region. In our analysis, the secondary airport is defined as a commercial airport

within 60 miles from the primary airport. A secondary airport is considered active if it has at

least 1% of the total enplanements in the metropolitan area.

0 1 2 3 4 5 6 7 8 9

JFK, EWR, LGAATL

ORD, MDWLAXDFW

BWI, IAD, DCAMIA, FLL

SFOIAHDENLASPHXMCODTWCLTBOSPHLMSPSEASLCTPASANCVGPDXSTLCLE

MEMPIT

No. of secondary airports

34 OEP

Airpo

rts in the CO

NUS

Active

Total

156

Fifteen multi-airport systems including 19 OEP airports and 25 active secondary airports are

identified based on 34 OEP airports in the NAS (see Figure 6-1). 59 inactive secondary airports

are located around 32 OEP airports. No secondary airport is located within 60 miles to ATL

(Atlantic) and SLC (Salt Lake City). Starting with 34 OEP airports in the Continental U.S., the

OEP airports within 60 miles to each other are grouped first. For example, JFK (Kennedy), EWR

(Newark) and LGA (LarGuardia) are grouped as one airport cluster. The secondary airports for

each OEP airport or OEP airport cluster are then identified from all the commercial airports in

the NAS. According to FAA airport categories, an airport is considered as a commercial airport

when it is served by scheduled flights and has a minimum of 2,500 enplanements in a calendar

year. According to this definition, a total 486 commercial airports are identified during 1990-

2008 based on the T-100 Domestic Segment data.

Figure 6-2: Evolution of Enplanements at Secondary (Active vs. Inactive) Airports in Boston

Area (Source: 1990-2008 T-100 Segment).

The traffic is more volatile at the inactive secondary airports compared with that at the active

secondary airports. The evolution of enplanements at five inactive airports in MAS of Boston is

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Enplan

emen

ts at A

ctive

Second

ary Airpo

rts in

Boston

Area (M

illions) PVD (Boston/Providence)

MHT (Boston/Manchester)

0

20

40

60

80

100

120

Enplan

emen

ts at Ina

ctive

Second

ary Airpo

rts in

Boston

Area (Tho

usan

ds) ORH (Boston/Worcester)

PSM (Boston/Portsmouth)

EWB (Boston/New Bedford)

BED (Boston/Hanscom Field)

PVC (Boston/Provincetown)

157

presented in Figure 6-2. As shown, the enplanements at the inactive airports ORH (Worcester)

and PSM (Portsmouth) fluctuated by large percentages over the period 1990-2008. The volatility

of traffic at ORH and PSM was caused by the interactions of few dominant carriers. The traffic

level and characteristics at the airports changed as the dominant airline(s) ceased or initiated

service. For example, ORH was served mainly by US Airways in early 1990s. The airport lost its

traffic as US Airways terminated service. The traffic then increased and decreased as Pan

American and Delta started and stopped service in early 2000s. Allegiant (a low-cost carrier)

then started service in 2005 and stopped service in 2006. In contrast, the traffic at two active

secondary airports PVD (Providence) and MHT (Manchester) had been on the rise since

Southwest Airlines entered these markets.

A threshold value of domestic Originating and Destination (O&D) passengers (or total

enplanements) in the metropolitan area that produce successful secondary airports was loosely

correlated with that suggested in the literature. The O&D passengers are those who either start

their itineraries in the metropolitan area or stop there for a reason other than changing aircraft to

their final destinations. The literature review (de Neufville, 1995) suggests that a threshold of 10

million O&D passengers in 1995 and 12 millions O&D passengers before 2010 is a factor for the

MAS to develop successfully. Another threshold of 18 million total enplanements in a

metropolitan area is presented by Bonnefoy and Hansman (Bonnefoy and Hansman, 2005). The

O&D passenger demand and the total enplanements for all the 15 MASs and 13 one-OEP airport

systems in 2008 are identified to examine such threshold for the success of a MAS.

Unfortunately, the transition between the MASs and the one-airport system was not apparent in

our analysis.

The catchment population share of the secondary airports did not increase from 1990 to 2000. It

is assumed that significantly more people live around secondary airports in MAS. Based on tract

Census data for 1990 and 2000, the catchment population of the secondary airports in each MAS

in 1990 is compared with that in 2000. The catchment population share of the secondary airports

in each MAS did not show increasing trends. The catchment population for each airport is

defined as the sum of populations of all the tracts to which the airport is the closest among all the

commercial airports available. The catchment population share of one airport is the share of the

airport’s catchment population over the sum of the catchment populations for all the airports in

158

one MAS. More years’ data may be necessary to approve this hypothesis. Therefore, more

analysis is recommended after Census 2010 data becomes available.

Figure 6-3: Evolution of Enplanements Share at Secondary Airports



Airlines decisions to expand service at the secondary airport will drive the traffic redistribution

among the airports in the MAS. It is based on expected profitability instead of accommodating

future air travel demand, and cannot be predicted with any degree of certainty (SH&E, 2009).

The traffic distribution among the airports in each of the 15 MASs studied is analyzed based on

T-100 data. Among 15 MASs, the share of passenger enplanements at secondary airports

increased at nine MASs in the period 1990-2008. The share of passenger enplanements at

secondary airports in the rest of MASs decreased in the same time period. The share of

enplanements of active secondary airports in seven MASs is higher than 18% in 2008. The

evolution of share of passenger enplanements at secondary airports in these seven MASs is

presented in Figure 6-3. The share of passenger enplanements is combined in the figure when

there is more than one secondary airport in a MAS.

The secondary airport share of passenger enplanements increased mostly in Boston area. As

presented in Figure 6-4, the total share of passenger enplanements at two active secondary

airports PVD (BOS/Providence) and MHT (BOS/Manchester) increased from 12% in 1990 to

25% in 2008. Southwest Airlines started operations at PVD in 1996 and MHT in 1998. This fact

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

Enplan

emen

ts Sha

re at

Second

ary Airpo

rts in the MAS

MIA

SFO

LAX

BOS

ORD

IAH

CVG

159

is considered the main drive for the substantial traffic growth at the Boston area secondary

airports.

Figure 6-4: Evolution of Enplanements Share at Active Airports in Boston Area


Figure 6-5: Evolution of Enplanements Share at Active Airports in Houston Area


On the other hand, the secondary airport share of passenger enplanements at Houston, Cleveland,

Tampa, Orlando and Philadelphia decreased over time. At HOU (Houston/Hobby) a reduction

from 35% in 1990 to 19% in 2008 was observed. The reduction in passenger enplanements is

caused by the lower traffic growth at HOU compared with the primary airport IAH

(Houston/Intercontinental). As shown in Figure 6-5, the total enplanements at IAH increased

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%Sh

are of Enp

lane

men

ts in

Multi‐Airpo

rt System of B

oston

BOS (Boston/Logan)

PVD (Boston/Providence)


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

Share of Enp

lane

men

ts in

Multi‐

Airpo

rt System of H

ouston

IAH (Houston/Intercontinental)

HOU (Houston/Hobby)

160

from 8.4 million in 1990 to 20 million in 2008 with an average 5% growth per year. In the same

period, the total enplanements at HOU remained around 5 million. FSCs covered 30% of total

enplanements at HOU in 1990. But they removed almost all their flights from HOU by 2008.

LCCs mainly represented by Southwest increased their enplanements from 3.2 million to 4.6

million in the same period. This growth rate (2%) is lower compared with that of FSCs at IAH.

Figure 6-6: Evolution of Enplanements Share at Active Airports in Miami Area


In the MAS of Miami and Tampa, the trend of the two secondary airports passenger

enplanements share is different. The enplanements share of one secondary airport increased

while the share at the other secondary airport in the same MAS decreased or stayed same. As

presented in Figure 6-6, the passenger enplanements share at FLL (Miami/Fort Lauderdale)

increased rapidly after 1995, but the share at the other secondary airport PBI (Miami/Palm

Beach) decreased from 14% to 11%. The fast traffic growth is due to starting operations of

several LCCs including Southwest, Spirit Airlines. FLL is focus city for Spirit, JetBlue, AirTran

and Allegiant. In contrast to FLL, the traffic growth at PBI was much lower. The evolutions of

passenger enplanements share at active airports in other MASs are included in Appendix E.1.

6.4 Markets served at secondary airports in MAS

An airport is considered as served if at least 365 non-stop flights are scheduled from the airport

in a MAS. For example, in the San Francesco bay area, SFO served 55 airports in 2008, while

only 25 and 23 airports are served at OAK (Oakland) and SJC (San Jose), respectively. The

0%

10%

20%

30%

40%

50%

60%

70%

80%

Share of Enp

lane

men

ts in

Multi‐

Airpo

rt System of M

iami

MIA (Miami/Int'l)

FLL (Miami/Fort Lauderdale)

PBI (Miami/Palm Beach)

161

airports served at SFO, OAK and SJC in the multi-airport system of San Francesco in 2008 are

shown in Figure 6-7. The same figures for some other MASs are also included in Appendix E.2.

In all the figures, the main hub and the secondary hub airports for FSCs serving an airport are

highlighted with red dots and red circles, respectively. The main hub and secondary hub airports

for LCCs serving an airport are highlighted in cyan dots and cyan circles. Green lines represent

the markets served exclusively by FSCs exclusively. Blue lines represent markets served by

LCCs exclusively. Magenta represents the market served by both FSCs and LCCs.

Figure 6-7: Markets Served at Airports in Multi-Airport System of San Francesco

(Source: 2008 OAG).

In the analysis, we found that the number of markets served by secondary airports is less than

that at the primary airport in the same metropolitan area. The secondary airports tend to serve the

short- or medium-haul high density markets in large metropolitan areas. The airports serving the

same metropolitan area are combined into one market. For example, all five commercial airports

in the Los Angeles area are served in the MAS of San Francesco. These five airports are

SFO (Year = 2008)

SJC (Year = 2008)OAK (Year = 2008)

162

combined as one market. A total of 54 markets are served at three airports in the San Francesco

bay area. SFO, OAK and SJC airport serve 50, 21 and 18 markets, respectively. OAK and SJC

together serve 25 of 50 airports. 17 airports served by both OAK and SJC overlap with the

markets served by SFO. The top 21 markets receiving 90% of total seats supplied in the San

Francesco MAS are presented in Figure 6-8. The markets are presented in decreasing order of the

seats supplied from the bottom to the top. The top five are short- or medium-haul markets

including Los Angeles (LAX, BUR, SNA, ONT, LGB), San Diego (SAN), Las Vegas (LAS),

Seattle (SEA), Phoenix (PHX). They are also the top five markets served by both OAK and SJC.

The MAS supplies the seats most to the short-haul market of Los Angeles (LAX, BUR, SNA,

ONT, LGB). 35% and 34% of the total seats at OAK and SJC, respectively are supplied to Los

Angeles. 82% and 79% of the seats at OAK and SJC are supplied to eight short- or medium-haul

markets including the top five markets and Denver (DEN), Portland (PDX) and Salt Lake City

(SLC) in decreasing order. The share of seats for the same eight markets at SFO is much lower

(46%).

Figure 6-8: Seat capacity by Market Served at Airports in San Francesco Bay Area

(Source: 2008 OAG).

0 2 4 6 8

LAX,BUR,SNA,ONT,LGBSANLASSEAPHX

JFK,EWRDEN

ORD,MDWPDXDFWHNLSLC

IAH,HOUATLIADRNOMSPBOSPHLCLTBOI

Seats Capacity (Millions)

SFOOAKSJC

0.0

0.5

1.0

1.5

2.0

2.5

LAX BUR SNA ONT LGB

Seats (Millions)

0.0

0.5

1.0

1.5

JFK EWR

Seats (Millions)

0.0

0.2

0.4

0.6

0.8

1.0

ORD MDW

Seats (Millions)

0.0

0.1

0.2

0.3

0.4

IAH HOU

Seats (Millions)

163

The secondary airport share of seats for the short- or medium-haul markets is higher than the

average level observed at primary airports. The secondary airport share of seats is lower than the

average level observed at primary airport for the long-haul markets. This is especially evident for

transcontinental markets. Among the top 21 markets covering 90% of the total seats supplied in

the San Francesco bay area MAS, seven of them are short-haul markets, three of them are

medium-haul markets, and the rest (i.e., 11) are long-haul markets. Two secondary airports

together cover 46% of total seats supplied in the San Francisco MAS. The secondary airport

share of seats for all ten short- or medium-haul markets is higher than 46%, while it is lower than

46% for all the 11 long-haul markets. Additionally, it is lower than 20% for the transcontinental

long-haul markets. For example, New York City (JFK, EWR) has a 6% market share, 14% for

IAD (Washington/Dulles) market. The secondary airports do not serve three transcontinental

markets: BOS (Boston/Logan), PHL (Philadelphia) and CLT (Charlotte).

Airlines provide service at multiple airports in the same metropolitan area. FSCs focus their

service mainly on the primary airport that is a hub. LCCs often focus their service on the

secondary airports, but sometimes they also select primary airports as focus city. The seat

distribution of LCCs among the airports at one metropolitan area is more spread than for FSCs.

Using economies of scale, airlines are likely to increase their frequency share in a market in

order to gain market share. Therefore, the airlines typically tend to concentrate their services at

one airport even though there are more airports in the same metropolitan area. This prevents the

development of multi-airport system. On the other side, there is a possibility to increase market

share by adding flights at the primary airport when the frequency reaches a certain threshold at

the primary airport. Airlines can then benefit from serving more in the same metropolitan area.

For example, ten carriers including seven FSCs and three LCCs serve the San Francesco bay

area. The number of seats supplied by the ten carriers at each of three airports is presented in

Figure 6-9. FSCs focus their service on the primary airport SFO. However, they also serve at

least one secondary airport. Specifically, United, Alaska, US airways and Delta serve all the

three airports, and American, Continental and Northwest serve SFO and SJC only. Southwest

focuses mainly on OAK and Virgin America only serves SFO. Southwest and JetBlue both serve

all the three airports. The share of seats for the best served airport for Southwest and JetBlue is

53% and 55%. The share of seats for the best served airport in the San Francisco area is 68% (US

Airways) and 92% (United) for FSCs.

164

Figure 6-9: Seat capacity by Carrier at Airports in San Francesco Bay Area

(Source: 2008 OAG).

6.5 Traffic Distribution by Markets and Carrier at the Secondary Airports in MAS

Many of the secondary airports are dominated by a LCC. The share of seats supplied by the

LCCs at the secondary airports has increased during the period 1990-2008. Figure 6-10 presents

the LCCs share of enplanements at 44 airports in 15 multi-airport systems for 1990, 2000 and

2008. In 1990, Southwest was the only carrier to provide LCC service, and only six secondary

airports in five metropolitan areas were served with significant seats share: MDW (Chicago

Midway, 25%), ONT (Los Angeles/Ontario, 18%), BUR (Los Angeles/Burbank, 19%), OAK

(San Francesco/Oakland, 24%), HOU (Houston/Hobby, 71%) and DAL (Dallas/Love Field,

100%). In 2008, the LCCs share of enplanements at all 25 secondary airports is greater than

18%. It is no less than 40% at 22 secondary airports, and greater than 70% at ten secondary

airports. In contrast, the LCCs share of passenger enplanements is less than 40% at 17 of 19

primary airports.

0

2

4

6

8

10

12

14

Seats C

apacity (M

illions)

SJC (San Francesco/San Jose)

OAK (San Francesco/Oakland)

SFO (San Francesco/Int'l)

165

Figure 6-10: LCCs Share of Enplanements at 44 Airports in 15 Multi-Airport Systems in the

NAS (Source: 1990, 2000, 2008 T-100 Segment).

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

JFK (New York/Kenndy)EWR (New York/Newark)

LGA (New York/LaGuardia)ISP (New York/Islip)

HPN (New York/Westchester)SWF (New York/Stewart)ORD (Chicago/O'Hare)

MDW (Chicago/Midway)LAX (Los Angeles/Int'l)

SNA (Los Angeles/Orange County)ONT (Los Angeles/Ontario)BUR (Los Angeles/Burbank)

LGB (Los Angeles/Long Beach)DFW (Dallas/Fort Worth)DAL (Dallas/Love Field)

IAD (Washington/Dulles)BWI (Washington/Baltimore)

DCA (Wasington/Reagan)MIA (Miami/Int'l)

FLL (Miami/Fort Lauderdale)PBI (Miami/Palm Beach)SFO (San Francesco/Int'l)

OAK (San Francesco/Oakland)SJC (San Francesco/San Jose)

IAH (Houston/Intercontinental)HOU (Houston/Hobby)

MCO (Orlando/Int'l)SFB (Orlando/Sanford)

DAB (Orlando/Daytona Beach)DTW (Detroit/Metropolitan)

FNT (Detroit/Bishop)BOS (Boston/Logan)

PVD (Boston/Providence)MHT (Boston/Manchester)

PHL (Philadelphia/Int'l)ACY (Philadelphia/Atlantic City)ABE (Philadelphia/Lehigh Valley)

TPA (Tampa/Int'l)SRQ (Tampa/Sarasota)

PIE (Tampa/St. Petersburg)CVG (Cincinnati/Int'l)

DAY (Cincinnati/Dayton)CLE (Cleveland/Hopkins)

CAK (Cleveland/Akron‐Canton)

LCCs' Enplanements/Total Enplanements

44 Airpo

rts in 15

Multi‐Airpo

rt Systems in the NAS 2008

2000

1990

166

Figure 6-11: “Southwest Effect” at PVD (Boston/Providence) and MHT (Boston/Manchester)


The growth of traffic at the secondary airports is not only directly due to the fast growth of

LCCs. The traffic by FSCs at the secondary airport was also stimulated by the entry of LCCs.

This phenomenon is known as the “Southwest effect”. “Southwest effect” is defined as the

significant growth of enplanements and the reduction in average fares at airports that Southwest

enters, and/or the airports serving the same metropolitan area (Windle and Dresner, 1995 b;

Windle et al., 1996; Southwest Airlines, 1999). The Southwest effect in multi-airport system was

examined by Mc Kenna (1996). The Southwest effect is responsible for the increase in

enplanements at PVD and MHT as observed in Figure 6-11. PVD was served by all big six FSCs

before Southwest entered the market. The total enplanements at PVD had been around 1 million

during 1990-2005. The southwest effect is also observed at MHT airport in the same MAS. MHT

was served by three FSCs before Southwest entered the market in 1998. The total enplanements

increased from 0.3 million in 1990 to 0.4 million in 1997. The traffic served by FSCs and their

affiliated regional carriers grew quickly to 1.0 million during 1998-2000 following the entry of

Southwest.

0.0

0.5

1.0

1.5

2.0

Enplan

emen

ts (M

illions) PVD (Boston/Providence)

LCCs

FSCs and others

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Enplan

emen

ts (M

illions) MHT (Boston/Manchester)

LCCs

FSCs and others

167

Figure 6-12: Markets Served by Southwest at Airports in MAS of Los Angeles & Boston

(Source: 2008 OAG).

Because of LCCs point-to-point business model, Southwest tends to serve short- or medium-haul

high density markets. The Los Angeles and Boston MAS present different cases about how

Southwest Airlines selects markets at airports in an MAS. In the Los Angeles MAS, Southwest

selects the primary airport LAX as focus city. It also serves all the active secondary airports

except LGB (Long Beach) and is on top of all the other carriers in terms of total seats supplied at

all the airports in the Los Angeles area. 31% of the total seats in Los Angeles area are provided

by Southwest. This is equivalent to the number of seats supplied by United and American

Airlines. As presented in Figure 6-12, Southwest serves 18 markets at LAX, but only five high

density short-haul markets: OAK (Oakland/San Francesco, CA), SJC (San Jose/San Francesco,

CA) LAS (Las Vegas, NV), PHX (Phoenix, AZ) and SMF (Sacramento, CA) are mainly served

at the three secondary airports. 78% of seats provided by Southwest in Los Angeles area are for

these five markets. Southwest serves SFO from LAX, but does not serve it from any of the

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

Dep

artures S

ched

uled

Non‐stop Markets Served by Southwest

LAX (Int'l)*

SNA (Orange County)

ONT (Ontario)

BUR (Burbank)

LGB (Long Beach)

0

1,000

2,000

3,000

4,000

5,000

BWI MCO PHL MDW TPA FLL LAS PHX BNADep

artures S

ched

uled


BOS (Logan)

PVD (Providence)

MHT (Manchester)

168

secondary airports. In Boston MAS, Southwest serves two secondary airports PVD (Providence)

and MHT (Manchester). It entered PVD in 1996 and MHT in 1998. The markets served at PVD

and MHT by Southwest in decreasing order of seats supplied are presented in Figure 6-12.

Except for BNA (Nashville, TN) that is served only by PVD, all the other eight markets served

by Southwest at PVD and MHT overlap. Six of the eight markets are Southwest focus cities.

Additionally, Southwest service is evenly distributed between PVD and MHT for the eight

markets.


usually functions as one or more FSCs hub, FSCs concentrate their service mainly on the

primary airport. They tend to provide short- or medium-haul flights from the secondary airport to

their hub or secondary hub when they supply service at the secondary airport. For example, in

the Boston MAS, US Airways use Logan as a secondary hub. The markets served by US

Airways are presented in Figure 6-13. Only short- and medium-haul flights to its hubs are

supplied at PVD (Providence) and MHT (Manchester). As presented in Figure 6-13, the five

markets United serves from BOS are its five main hubs ORD (Chicago/Int’l), SFO (San

Francesco/Int’l), IAD (Washington/Dulles), DEN (Denver) and LAX (Los Angeles/Int’l) in

decreasing order of seats supplied by United. Only the short-haul flight to IAD and medium-haul

flight to ORD are supplied at PVD and MHT. The short-haul flights to IAD from PVD and MHT

are all operated with regional jets contrasting with the 30% of flights operated with regional jets

at BOS.

‐

1,000

2,000

3,000

4,000

5,000

6,000

Dep

artures Sche

duled

Non‐stop Markets Served by US Airways

BOS (Boston/Logan)



169

Figure 6-13: Markets Served by Full Service Carriers at Airports in Boston Area

(Source: 2008 OAG).

6.6 Aircraft Type at the Secondary Airports in MAS

The average seating capacity per aircraft at the secondary airports is higher than that of primary

airports in most of the MASs. This can be explained because secondary airports are dominated

by LCCs. LCCs typically use narrow-body jets with seat capacity between 100 and 150, i.e.,

Boeing 737 with 122 or 137 seats for Southwest airlines; Airbus A320 with 150 seats and

Embraer 190 with 100 seats for JetBlue, while FSCs dominating primary airports use affiliated

regional jets with less than 100 seats to feed their hubs.

As presented in Figure 6-14, the share of flights served by FSCs with turboprop and regional jets

with 37-99 seats at the secondary airports are higher than that of primary airports in 2008. The

impact of regional jets on the evolution of U.S. domestic airline route networks was of

considerable growth in the past decade (Oster and Strong, 2006). Regional jets have become a

significant component of air transport in the United States. Most of the increase in both aircraft

departures and seat capacity between 1995 and 2000 was due to affiliated regional jets (Oster

and Strong, 2006). Between May 1997 and October 2000, the number of regional jet departures

grew by 735% in contrast with 9% for mainline jets (GAO, 2001). To reflect the reduced

demand, the total seat capacity at the hub airports was lower in 2003, but the seats offered by

regional jets had increased at the same time.

No official definition of Regional Jet (RJ) exists in the airline industry (General Accounting

Office, 2001; DOT, 1998). Oster and Strong (2006) consider RJs as jet aircraft with 30 to 70

0

500

1,000

1,500

2,000

2,500

3,000

3,500

ORD SFO IAD DEN LAXDep

artures S

ched

uled

Non‐stop Markets Served by United

BOS (Boston/Logan)



170

seats and operated in schedule service by mainline airlines’ affiliated regional airlines. While the

most widely accepted definition considers RJs as all jet aircraft with less than 100 seats (Wong,

Pitfield and Humphreys, 2005). In contrast to turboprops, regional jets have higher level of

comfort, greater range, and higher speed. Turboprops were slower and had a cabin environment

that is noisier and subjected to more vibration than mainline jets. The range segment flown by

turboprops in the U.S. is typically from 125 to 150 miles. RJs operate at similar speeds and

altitude to narrow-body mainline jets and have as good cabin environment in terms of noise and

vibration. The range of RJs is typically from 350 to 1200 miles. Therefore RJs are a much better

substitute for narrow-body mainline jets in thin routes.

171

Figure 6-14: Distribution of Aircraft Type at 44 Airports in 15 Multi-Airport Systems in the

NAS (Source: 2008 OAG).

0% 50% 100%





















Aircraft Category of FSCs in 2008

Airpo

rts in Multi‐Airpo

rt System

P (7‐9)H (8)T (19‐74)JN (37‐99)JN (100‐252)JW (165 ‐ 397)

172

The regional jets have been used to replace turboprops, to supplement turboprops, to replace

mainline jets, to supplement mainline jets, and to initiate new service (DOT, 1998). Of all the

markets that received regional jet service since 1997, most were markets between the air carriers’

hub and spoke cities that were previously served with either turboprops or mainline jets; 41%

were completely new markets (DOT, 1998). The regional jets can be used to improve frequency

of service in mainline markets during times of day/days of week when the demand is not

sufficient to support mainline jet service. Use of RJs allows airlines to serve more distant cities

that previously were beyond the turboprops’ operating range and did not have sufficient

passengers to support the mainline jet service. The mainline legacy carriers have been using

affiliated regional jets to add new spokes to their hubs (Oster and Strong, 2006). For example,

Delta Air Lines initiated routes between Cincinnati and 36 airports within 1,000 miles to

Cincinnati. 34 of the new routes were served by its affiliated regional airline Comair using RJs

(Savage and Scott, 2004). Today, the story has reversed, before Delta merged with Northwest

airlines commercial traffic at Cincinnati peaked at 11.2 million passengers enplaned in 2005.

Since then traffic has rapidly declined to 5.1 million in the year 2009 (FAA 2010). This change

of fortunes can be explained by the consolidation of services of one carrier (i.e., Delta-

Northwest) into its Detroit hub.

6.7 Flight Distance at the Secondary Airports in MAS

Most long-haul flights are operated from primary airports that are usually main hubs or

secondary hubs for FSCs. In hub-spoke networks, the demand from various origins is combined

at the hub airport. Therefore, the condition of a sufficient demand for a long-haul flight can be

met more likely at the hub airport (Maertens, 2009). Long-haul flights requires better airport

infrastructure than their short- and medium-haul counterparts. All 25 active secondary airports

studied can handle the long-haul flights. The runway length at the 25 active secondary airports

ranges from 5,700 feet at SNA (Los Angeles/Orange County) to 12,200 feet at ONT (Los

Angeles/Ontario). Considering that 26% of total operations are long-haul flights at SNA and

transcontinental flights are also scheduled at the airport by Continental to its EWR hub, it is safe

to say that most of the active secondary airports can handle the long-haul flights.

173

Figure 6-15: Distribution of Flight Range at 44 Airports in 15 Multi-Airport Systems in the NAS

(Source: 2008 OAG).

0% 20% 40% 60% 80% 100%



HPN (New York/Westchester)SWF (New York/Stewart)

ORD (Chicago/O'Hare)MDW (Chicago/Midway)LAX (Los Angeles/Int'l)

















Distribution of Flight Range in 2008

Airpo

rts in Multi‐Airpo

rt System

Short‐haul (<=600)

Medium‐haul (600‐1200)

Long‐haul (>1200)

174

An observed trend is that most of the secondary airports do not capture long-haul flights

although all the 25 secondary airports meet the runway length requirement of long-haul flights.

As presented in Figure 6-15, most of the flights served at the secondary airports are short-haul

(no longer than 600 miles) or medium-haul (600-1200 miles) flights. In most of the multi-airport

systems, the share of long-haul flights at the secondary airports is less than that at the primary

airport. Consistently, the average flight distance at the secondary airports is lower than that at the

primary airport for almost all the multi-airport systems. For example, in the San Francesco bay

area multi-airport system, the average flight distance at OAK (Oakland) and SJC (San Jose) is

less than 650 miles, while the average for SFO is 1,014 miles. Compared with the average flight

distance of 750 miles at IAH (Houston/Intercontinental), the average flight distance at HOU is

563 miles. One exception is MDW (Chicago/Midway). The average flight distance (821 miles) at

secondary airport MDW (Chicago/Midway) is longer than that recorded at Chicago/O’Hare -

ORD (653 miles). The share of long-haul flights at MDW is 21.6% compared with 11.7% at

ORD. This anomaly could be explained because all the long-haul markets served by the FSCs at

ORD are also served by Southwest at MDW. ORD has more feeder traffic from the nearby spoke

airports. Another exception is LGB (Los Angeles/Long Beach) whose longest runway length is

10,000 feet. The average flight distance at this airport is 912 miles, which is longer than that at

other three secondary airports in Los Angeles. LGB is a hub for JetBlue. JetBlue covers 67% of

the seats supplied at LGB and provides transcontinental flights to its three other hub airports on

the east coast: JFK (New York/Kennedy), IAD (Washington/Dulles) and BOS (Boston/Logan).

Restricted by the Wright and Shelby Amendments, 96% of flights operated at the secondary

airport DAL (Dallas/Love Field) in the Dallas MAS are short-haul flights. DAL average flight

distance is only 344 miles. The current primary airport DFW (Dallas/Fort Worth) was officially

opened in 1974 to replace DAL (Dallas/Love Field). DAL is one of Southwest Airlines focus

cities. It reemerged as the secondary airport because of Southwest’s success. To protect DFW,

the markets served at DAL are restricted by Wright and Shelby Amendments that prevents the

airlines from serving passenger flights to other states except four neighboring states (Louisiana,

Arkansas, Oklahoma and New Mexico), and four other states (Kansas, Mississippi, Alabama and

Missouri). This prescription applies to aircraft with 56 or more seats, or aircraft weighting

300,000 or more pounds. Due to Wright and Shelby Amendments, 15 short- or medium- haul

markets are flown from DAL by Southwest. The airline has 96% share of seats at the airport.

175

6.8 Connectivity at the Secondary Airports in MAS

The secondary airports mainly serve the domestic Originating and Destination (O&D)

passengers who either start their itineraries in the metropolitan area or stop there for a reason

other than changing aircraft to their final destinations. Connecting passengers prefer the primary

airport that is usually a FSC hub airport because it supplies easier and more flexible connection

possibilities. The share of connecting passengers over total passengers at all 25 active secondary

airports and 19 primary airports is estimated based on the latest (2008) DB1B data. Except

MDW (Chicago Midway), HOU (Houston/Hobby) and DAL (Dallas/Love Filed) that have a

significant share of connecting passengers, all the secondary airports serve O&D passengers. As

shown in Figure 6-16, the share of connecting passengers at most of the secondary airports is less

than 5%, while the share is 30% at MDW, 27% at DAL and 23% at HOU. These three secondary

airports are Southwest focus cities. Southwest seats share is 89% at MDW, 96% at DAL and

93% at HOU. Southwest has a much smaller share of connecting passengers compared with

FSCs (50%-70%). About 25% of Southwest passengers made connections since 1998 (Oster and

Strong, 2006).

176

Figure 6-16: Connecting Rate at 44 Airports in 15 Multi-Airport Systems in the NAS

(Source: 2008 Domestic DB1B).

0% 10% 20% 30% 40% 50% 60% 70% 80%





















Connecting Rate in 2008

44 Airpo

rts in 15

Multi‐Airpo

rt Systems in the NAS

Red bar ‐ Primary airportBlue bar ‐ Secondary airport

177

The markets covering at least 90% of the connecting passengers at MDW and HOU are

presented in Figure 6-17. 90% of the connecting passengers at MDW are originating from or

destined to 24 markets including 12 short-haul markets, and medium- and long-haul markets on

the East and West coasts; 92% of the connecting passengers at HOU are originating from or

destined to 10 markets including eight short-haul markets and three medium-haul markets in

Florida: MCO (Orlando/Int’l), TPA (Tampa/Int’l), FLL (Miami/Fort-Lauderdale).

Figure 6-17: Connecting Markets Covering 90% of Connecting Passengers at MDW and HOU


The top five markets with most connecting passengers at MDW include three short-haul markets

DTW (Detroit, MI), IAD (Washington/Dulles), CLE (Cleveland, OH), and two medium-haul

markets MHT (Boston/Manchester, MA) and PHL (Philadelphia, PA) in decreasing order of

connecting passengers at MDW. 39% of passengers connecting at MDW are originating from or

destined to the top five connecting markets. The top five markets covering 69% connecting

passengers at HOU are within 302 miles from HOU. These markets are: HRL (Harlingen, TX),

MSY (New Orleans, LA), AUS (Austin, TX), CRP (Corpus Christi, TX) and SAT (San Antonio,

TX) in decreasing order of connecting passengers at HOU. 20% of connecting passengers at

HOU are originating from or destined to HRL (Harlingen, TX).

The medium- and long-haul connecting markets for MDW and HOU are located in big

metropolitan areas. Most of them are Southwest’s focus cities. The top five medium- and long-

haul markets with a large amount of connecting passengers are:

For MDW:

178

- Medium-haul (600-1200 miles): MHT (Boston/Manchester, MA), PHL (Philadelphia,

PA), DEN (Denver, CO), PVD (Boston/Providence, MA) and ISP (New York/Islip, NY)

- Long-haul (1,200 or more miles): PHX (Phoenix, AZ), SEA (Seattle, WA) OAK (San

Francesco/Oakland, CA), LAS (Las Vegas) and LAX (Los Angeles/Int’l, CA)

For HOU:

- Medium-haul (600-1200 miles): MCO (Orlando/Int’l, FL), TPA (Tampa/Int’l, FL), FLL

(Miami/Fort-Lauderdale, FL), BNA (Nashville, TN) and ELP (El Paso, TX)

- Long-haul (1,200 miles or more): LAX (Los Angles), BWI (Baltimore, MD) and LAS

(Las Vegas, NV), OAK (San Francesco/Oakland, CA) and SAN (San Diego, CA)

In 2008, 97% of the passengers connecting at MDW, DAL and HOU had only one stopover at

the connecting airport to travel between their origin and destination. The great circle distance

between top 20 OD airport-pairs of connecting passengers at MDW ranges from 632 miles for

DTW (Detroit, MI)-MCI (Indianapolis, IN) to 2,518 miles for PHL (Philadelphia, PA)-SFO (San

Francesco/Int’l, CA); The great circle distance between the top 20 OD airport-pairs of

connecting passengers at HOU ranges from 425 miles for AUS (Austin, TX)-HRL (Harlingen,

TX) to 1,664 miles for LAX (Los Angeles/Int’l, CA)-HRL (Harlingen, TX); The OD pairs

covering 90% of total connecting passengers at MDW and HOU are shown in Figure 6-18.

Figure 6-18: OD Airport-Pairs Covering 90% of Connecting Passengers at MDW and HOU


As presented in Appendix E.5, the top five OD pairs with most connecting passengers are:

179

For MDW:

- DTW (Detroit, MI) - MCI (Harlingen, TX)

- PIT (Pittsburgh, PA) - STL (St. Louis, MO)

- BNA (Nashville, TN) - OMA (Omaha, NE)

- PHX (Phoenix, AZ) - IAD (Washington/Dulles)

- DEN (Denver, CO) - MHT (Boston/Manchester, MA)

For HOU:

- AUS (Austin, TX) - HRL (Harlingen, TX)

- DAL (Dallas, TX) - MSY (New Orleans, LA)

- AUS (Austin, TX) - MSY (New Orleans, LA)

- HRL (Harlingen, TX) - MDW (Chicago/Midway, IL)

- AUS (Austin, TX) - MCO (Orlando/Int’l, FL)

6.9 Conclusion

Fifteen multi-airport systems including 19 OEP airports and 25 active secondary airports were

identified in the NAS for this study. 59 inactive secondary airports are located around 32 OEP

airports. No secondary airport is located within 60 miles to ATL (Atlantic) and SLC (Salt Lake

City). The traffic is more volatile at the inactive secondary airports compared with that at the

active secondary airports. A threshold value of the domestic Originating and Destination (O&D)

passengers (or total enplanements) in the metropolitan area that produce successful secondary

airports was loosely correlated with that suggested in the literature. This threshold value has been

suggested to be 10 million enplanements. The catchment population share of the secondary

airports did not increase significantly from 1990 to 2000. More years of data may be necessary to

validate this hypothesis. Therefore, more analysis is suggested using Census 2010 when the data

becomes available.


Among 15 MASs in the analysis, the enplanements share of secondary airports increased in nine

MASs during 1990-2008. The secondary airport enplanements share decreased in MAS of

Houston, Cleveland, Tampa, Orlando and Philadelphia. In the Miami and Tampa MAS, the trend

of the two secondary airports’ enplanements share is different. The enplanements share of one

180

secondary airport increased while the share at the other secondary airport in the same MAS

decreased or stayed same. The enplanements share of active secondary airports in seven MASs is

higher than 18% in 2008.

The number of markets served at the secondary airports is less than that at the primary airport in

the same metropolitan area. The secondary airports tend to serve the short- or medium-haul high

density markets in large metropolitan areas. The secondary airport seats share for the short- or

medium-haul markets is higher than the average seats offered at the primary airport. The

opposite trend is shown on long-haul markets, especially the transcontinental markets. Airlines

provide service at multiple airports in the same metropolitan area. FSCs focus their service

mainly on the primary airport (i.e., usually their hubs). LCCs often focus their service on the

secondary airports, but they also select primary airports as their focus city. The LCCs’ seats

distribution among the airports in one metropolitan area is more evenly distributed than FSCs.

Most of the secondary airports are currently dominated by the LCCs. The share of seats supplied

by the LCCs at the secondary airports has increased during the period 1990-2008. The growth of

traffic at some of the secondary airports is not only directly due to the fast growth of LCCs, but

also by the response adopted by FSCs while stimulated by the entry of LCCs. This phenomenon

is known as the “Southwest effect”.


usually functions as one or more FSCs’ hub, FSCs concentrate their service mainly on the

primary airport in all the MASs analyzed. They tend to provide short- or medium-haul flights

from the secondary airport(s) to their hub(s) or secondary hub(s) when they supply service at the

secondary airport.

The average seat capacity per aircraft at the secondary airports is higher than that of primary

airports in most of the MASs. The share of flights served by FSCs with turboprop and regional

jets with 37-99 seats at the secondary airports are higher than that of primary airports in 2008.

Most long-haul flights are operated at the primary airports that are usually main hubs or

secondary hubs for FSCs. Most of the secondary airports do not capture long-haul flights

although all the 25 secondary airports meet the runway length requirement of long-haul flights.

181

The secondary airports mainly serve the domestic O&D passengers. Except MDW (Chicago

Midway), HOU (Houston/Hobby) and DAL (Dallas/Love Filed) that has significant share of

connecting passengers, all the secondary airports mainly serve O&D passengers. The share of

connecting passengers at most of the secondary airports is less than 5%, while the share is 30%

at MDW, 27% at DAL and 23% at HOU.

6.10 Reference

Bonnefoy, P., (2004). Emergence and Impacts of the Emergence of Secondary Airports in the

United States. American Institute of Aeronautics and Astronautics, ATIO6497. Chicago, IL:

American Institute of Aeronautics and Astronautics.

Bonnefoy, P., (2008). Scalability of the Air Transportation System and Development of Multi-

Airport Systems: A Worldwide Perspective, Ph.D. dissertation, Massachusetts Institute of

Technology, Cambridge, MA.

Bonnefoy, P., & Hansman, R. J., (2005). Emergence of Secondary Airports and Dynamics of

Multi-Airport Systems. Cambridge, MA: Massachusetts Institute of Technology. available at:

dspace.mit.edu/handle/1721.1/34908

Bolgeri P., Dray L., Evans A. & Schäfer A., (2008), The Emergence of Multi-Airport Systems,

12th Air Transport Research Society (ATRS) World Conference, 6-10 July 200. Athens,

Greece.

Clinton V. Oster Jr, and John S. Strong., (2006). “The Evolution of U.S. Domestic Airline Route

Networks since 1990” Transportation Research Record: The Journal of the Transportation

Research Board, No. 1951.

Cohas, F.J., Belobaba, P.P. and Simpson, R. W., (1995). "Competitive fare and frequency effects

in airport market share modeling," Journal of Air Transport Management, vol. 2, no. 1, pp.

33-45, Mar. 1995.

de Neufville, R., (1995). Management of Multi-Airport Systems: A Development Strategy.

Journal of Air Transport Management Vol. 2, No. 2, pp. 99-110.

de Neufville, R., and Odoni, A., (2003). Airport systems: Planning, Design, and Management,

McGraw-Hill, New York.

de Neufville, R., (2004). Multi-Airport Systems in the Era of No-Frills Airlines. Transportation

Research Board conference, (pp. 1-19). Washington, DC.

182

de Neufville, R., (2005). The Future of Secondary Airports: Nodes of a parallel air transport

network? English version of article prepared for the journal Cahiers Scientifiques du

Transport Cahiers Scientifiques du Transport, Issue 47, pp. 11-38.

Doganis, R., (2001). The Airline Business in the 21st Century. Routledge, London

Ishii, J., Jun, S., & Van Dender, K. (2009). Air Travel Choices in Multi-Airport Markets. Journal

of Urban Economics Vol. 65, No. 2, pp. 216-227.

DOT, 1998. Profile: regional jets and their emerging roles in the US aviation market. Office of

the assistant secretary for aviation and international affairs, Department of Transportation,

Washington, DC, June.

GAO, 2001. Regional Jet Service Yet to Reach Many Small Communities.General Accounting

Office, Washington, DC, February.

Hess, S., Polak, J.W., (2006). Airport, airline and access mode choice in the San Fransico Bay

area. Papers in regional Science 85 (4), 543-567.

Maertens, S. (2009). “Drivers of long haul flight supply at secondary airports in Europe”,

Journal of Air Transport Management, Volume 16, Issue 5, September 2010, Pages 239-243

McKenna, J., (1996). Carriers in Florida brace for Southwest. Aviation Week & Space

Technology. January 22.

MITRE, (2000). Regional Jets. The MITRE Corporation (<www.mitre.org> accessed 28/05/02).

Oster, C.V., Jr., and Strong, J.S., (2006). Evolution of U. S. Domestic Airline Route Networks

Since 1990. In Transportation Research Record: Journal of the Transportation Research

Board, No. 1951, Transportation Research Board of the National Academies, Washington, D.

C., 2006, 52-59.

Pels, E., Nijkamp, P., Rietveld, P., (2003). Access to and competition between airports: a case

study for the San Francisco Bay Area. Transportation Research A 37 (1), 71-83.

Savage, I. and Scott, B., (2004). Deploying Regional Jets to Add New Spokes to a Hub. Journal

of Air Transport Management, Vol. 10, 2004, 147-150.

SH&E, (2009). Alternative Strategies for Accommodating Future Aviation Demand

(www.mtc.ca.gov/planning/air_plan/RAPC_11-20-09.ppt).

Tien, S.-L. Schonfeld, P., (2007). “Passenger Market Equilibrium for Competing Airports in

Multiple Airport Region” Transportation Research Record: Journal of the Transportation

Research Board, No. 2007, pp. 13-21.

183

Warnock-Smith, D. & Potter, A., (2005). An exploratory study into airport choice factors for

European low-cost airlines. Journal of Air Transport Management Vol. 11, No. 6, pp. 388-

392.

Windle, R. and Dresner, M., (1995 a). "Airport Choice in a Multiple Airport Region", ASCE

Journal of Transportation Engineering, Vol. 21(4), July/August 1995, pp. 332-337.

Windle, R., Dresner, M., (1995 b). The short and long run effects of entry on US domestic air

routes. Transportation Journal 35 (2), 14–25.

Windle, R., Lin, J., Dresner, M., (1996). The impact of low-cost carriers on airport and route

competition. Journal of Transport Economics and Policy XXX (3), 309–328.

Wong, D.K.Y., Pitfield, D.E., Humphreys, I.M., (2005). The impact of regional jets on air

service at selected US airports and markets. Journal of Transport Geography, 13 (2), 151-

163.

184

7 Appendix

185

Appendix A:

An Agent-Based Model of Airline Evolution

186

A.1 Evolution of Airline Profit, Load Factor, Airfare, Aircraft Size, Market Share for the 1st

Scenario (Adjust Airfare and Aircraft size with Fuel Price: $1.5/gal)

0 200 400 600 800 1000-4

-3

-2

-1

0

1

2x 10

6

Generation

Airl

ine

Tota

l Pro

fit

Evolution: Airfare & Aircraft Size of each Flight / Fuel - $1.5/gal

0 200 400 600 800 10000.5

0.6

0.7

0.8

0.9

1

Generation

Airl

ine

Avg

. Loa

d Fa

ctor

Evolution: Airfare & Aircraft Size of each Flight / Fuel - $1.5/gal

0 200 400 600 800 1000100

150

200

250

300

350

Generation

Airl

ine

Avg

. Airf

are

per F

light

Evolution: Airfare & Aircraft Size of each Flight - $1.5/gal

187

0 200 400 600 800 1000100

150

200

250

300

Generation

Airl

ine

Avg

. Airc

raft

Siz

e: S

eats

Evolution: Airfare, Aircraft Size of each Flight / Fuel - $1.5/gal

0 200 400 600 800 10000.2

0.22

0.24

0.26

0.28

0.3

0.32

Generation

Agg

rega

te M

arke

t Sha

re o

f CA

Evolution: Airfare and Aircraft Size of each Flight / Fuel - $1.5/gal

0 200 400 600 800 10000.25

0.3

0.35

0.4

Generation

Mar

ket S

hare

of e

ach

Airl

ine


188

A.2 Evolution of Airline Profit, Load Factor, Airfare, Aircraft Size, Market Share for the 2nd

Scenario (Adjust Airfare and Aircraft size with Fuel Price: $2.0/gal)

0 200 400 600 800 1000-4

-3

-2

-1

0

1

2x 10

6

Generation

Airl

ine

Tota

l Pro

fit


0 200 400 600 800 10000.5

0.6

0.7

0.8

0.9

1

Generation

Airl

ine

Avg

. Loa

d Fa

ctor


0 200 400 600 800 1000100

150

200

250

300

350

Generation

Airl

ine

Avg

. Airf

are

per F

light


189

0 200 400 600 800 1000100

150

200

250

300

Generation

Airl

ine

Avg

. Airc

raft

Siz

e: S

eats


0 200 400 600 800 10000.2

0.22

0.24

0.26

0.28

0.3

0.32

Generation

Agg

rega

te M

arke

t Sha

re o

f CA


0 200 400 600 800 10000.25

0.3

0.35

0.4

Generation

Mar

ket S

hare

of e

ach

Airl

ine


190

A.3 Evolution of Airline Profit, Load Factor, Airfare, Aircraft Size, Market Share for the 3rd

Scenario (Adjust Airfare, Aircraft size and Add/Cancel Flights with Fuel Price: $1.5/gal)

0 200 400 600 800 1000

-2

-1

0

1

2x 10

6

Generation

Airl

ine

Tota

l Pro

fitEvolution: Airfare, Aircraft Size of each Flight

& Open New Flights & Cancel Flights: Fuel - $1.5/gal

0 200 400 600 800 10000.5

0.6

0.7

0.8

0.9

1

Generation

Airl

ine

Avg

. Loa

d Fa

ctor

Evolution: Airfare, Aircraft Size of each Flight & Open New Flights & Cancel Flights: Fuel - $1.5/gal

0 200 400 600 800 1000100

150

200

250

300

Generation

Airl

ine

Avg

. Airc

raft

Siz

e: S

eats


191

0 200 400 600 800 1000100

150

200

250

300

350

Generation

Airl

ine

Avg

. Airf

are

per F

light


0 200 400 600 800 1000 1200

0.3

0.32

0.34

0.36

0.38

Generation

Mar

ket S

hare

of e

ach

Airl

ine


0 200 400 600 800 10000.2

0.22

0.24

0.26

0.28

0.3

0.32

Generation

Agg

rega

te M

arke

t Sha

re o

f CA

Evolution: Airfare, Aircraft Size of each Flight & Open New Flights & Cancel Flights / Fuel - $1.5/gal

192

A.4 Input Data for the Model

Table A-1: Distance (County – County): Miles.

Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 178 382 517 361 350 584 650 597 764 1036 934 1032 1298 1231 2 178 0 216 382 334 435 575 698 715 769 1050 1066 1098 1331 1320 3 382 216 0 194 324 523 533 712 811 724 1002 1167 1112 1296 1357 4 517 382 194 0 295 533 425 637 803 599 863 1149 1018 1161 1276 5 361 334 324 295 0 239 241 390 509 434 716 861 791 1000 1033 6 350 435 523 533 239 0 320 307 289 456 708 646 681 955 888 7 584 575 533 425 241 320 0 224 470 195 477 775 593 766 851 8 650 698 712 637 390 307 224 0 294 201 406 557 403 649 645 9 597 715 811 803 509 289 470 294 0 495 651 357 498 824 647 10 764 769 724 599 434 456 195 201 495 0 282 725 437 572 705 11 1036 1050 1002 863 716 708 477 406 651 282 0 752 318 298 558 12 934 1066 1167 1149 861 646 775 557 357 725 752 0 466 795 443 13 1032 1098 1112 1018 791 681 593 403 498 437 318 466 0 342 268 14 1298 1331 1296 1161 1000 955 766 649 824 572 298 795 342 0 438 15 1231 1320 1357 1276 1033 888 851 645 647 705 558 443 268 438 0

193

Table A-2: Distance between County and Airport: Miles.

County \ Airport 1 2 3 4 5 6 7 8 9 10

1 922 294 631 1086 1284 1197 107 548 247 805 2 994 338 563 1121 1330 1274 271 387 77 913 3 1019 406 446 1096 1311 1296 455 171 139 987 4 937 422 283 972 1187 1205 562 124 309 949 5 696 136 277 793 1005 974 349 399 309 669 6 572 117 467 742 936 849 271 629 452 478 7 513 299 203 563 778 780 546 549 546 560 8 306 365 425 436 636 585 578 757 691 339 9 387 404 666 627 779 630 496 908 739 210 10 383 471 331 374 590 622 712 722 741 506 11 346 742 582 144 337 458 975 985 1023 564 12 406 761 978 660 716 485 828 1259 1094 220 13 113 761 765 199 282 187 946 1141 1093 330 14 442 1009 880 213 98 344 1225 1283 1309 671 15 339 982 1030 417 342 102 1135 1398 1324 438

194

Table A-3: County – County Passenger Demand (Business; Income Group 1).

Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 275 238 118 165 286 331 107 172 190 190 182 80 161 324 2 158 0 219 161 213 186 98 133 156 265 196 241 335 125 50 3 130 181 0 278 127 95 244 81 86 225 305 287 84 281 178 4 126 320 303 0 216 246 193 262 108 147 255 46 132 88 139 5 209 265 157 218 0 57 79 334 146 230 60 268 166 127 321 6 137 84 182 152 65 0 196 261 270 38 143 91 305 236 104 7 155 45 59 162 244 110 0 183 194 95 157 90 47 294 42 8 279 223 116 240 324 168 43 0 101 159 304 260 239 100 290 9 117 341 135 198 285 264 90 32 0 166 240 40 84 111 289 10 304 327 337 174 132 38 176 277 287 0 273 233 150 227 174 11 56 330 65 86 116 195 38 212 111 164 0 304 337 134 156 12 71 110 121 184 181 158 66 199 127 122 307 0 266 75 91 13 59 135 146 288 186 226 35 165 305 136 279 189 0 266 128 14 248 204 129 326 45 53 199 271 270 317 218 249 289 0 178 15 245 159 322 329 322 104 55 276 134 78 108 176 88 203 0

195


Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 550 477 235 331 571 663 214 343 381 381 364 161 322 648 2 316 0 437 322 427 373 196 266 313 531 391 482 670 249 101 3 260 363 0 557 254 190 489 162 172 450 609 574 168 563 355 4 252 640 607 0 433 492 385 524 216 293 510 93 265 176 278 5 418 530 314 435 0 115 158 667 292 459 120 536 332 254 642 6 273 167 365 304 129 0 393 522 541 77 287 183 610 472 208 7 310 89 119 324 487 219 0 365 388 191 314 181 94 588 84 8 557 446 232 480 647 336 86 0 203 318 608 519 477 200 579 9 235 682 270 396 570 528 180 63 0 332 480 79 167 222 578 10 608 655 673 347 264 75 352 555 575 0 547 467 301 454 347 11 112 661 130 172 232 391 77 424 223 327 0 609 675 268 312 12 142 220 243 369 362 316 133 399 254 244 615 0 532 149 182 13 117 270 292 575 373 452 70 330 610 271 558 377 0 532 256 14 495 408 259 652 89 105 399 542 540 635 436 497 578 0 357 15 490 319 645 658 643 208 110 551 268 157 215 352 176 406 0

196


Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 595 515 254 358 617 716 231 371 412 411 393 174 348 701 2 342 0 473 348 461 403 212 287 338 574 423 521 724 269 109 3 281 392 0 602 274 205 528 175 186 486 658 620 181 608 384 4 273 692 656 0 468 532 416 566 233 317 551 100 286 190 301 5 452 573 339 470 0 124 171 721 315 496 130 579 359 275 694 6 295 181 394 328 140 0 424 565 584 83 310 197 659 510 224 7 335 96 128 350 527 237 0 395 419 206 340 196 102 636 91 8 602 482 250 519 700 363 93 0 219 344 657 561 516 217 626 9 254 737 292 427 616 571 194 68 0 359 519 86 181 239 625 10 657 708 727 375 285 81 380 600 621 0 591 504 325 491 375 11 121 714 140 186 251 422 83 458 241 354 0 658 729 290 337 12 154 238 262 399 391 341 144 431 274 264 664 0 574 161 197 13 127 292 315 622 403 489 76 357 659 293 603 407 0 575 277 14 535 440 280 704 96 114 431 585 583 686 472 537 624 0 385 15 530 344 697 711 695 224 119 596 289 170 232 381 190 439 0

197


Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 355 308 152 213 369 428 138 221 246 246 235 104 208 418 2 204 0 282 208 275 240 127 171 202 342 252 311 432 161 65 3 167 234 0 359 164 123 315 105 111 290 393 370 108 363 229 4 163 413 392 0 279 318 249 338 139 189 329 60 171 114 180 5 270 342 203 281 0 74 102 430 188 296 77 346 214 164 414 6 176 108 235 196 83 0 253 337 349 50 185 118 394 304 134 7 200 58 77 209 314 142 0 236 250 123 203 117 61 380 54 8 359 288 149 310 418 217 56 0 131 205 392 335 308 129 374 9 152 440 174 255 368 341 116 41 0 214 310 51 108 143 373 10 392 422 434 224 170 49 227 358 371 0 353 301 194 293 224 11 72 426 84 111 150 252 49 274 144 211 0 393 435 173 201 12 92 142 157 238 233 204 86 257 164 158 396 0 343 96 117 13 76 174 188 371 240 292 45 213 393 175 360 243 0 343 165 14 320 263 167 420 58 68 257 349 348 410 282 321 373 0 230 15 316 206 416 425 415 134 71 356 173 101 139 227 114 262 0

198


Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 284 246 121 171 295 342 110 177 197 196 188 83 166 335 2 163 0 226 166 220 192 101 137 161 274 202 249 346 129 52 3 134 187 0 287 131 98 252 84 89 232 314 296 87 290 183 4 130 330 313 0 223 254 199 270 111 151 263 48 137 91 144 5 216 273 162 225 0 59 82 344 151 237 62 277 171 131 331 6 141 86 188 157 67 0 203 270 279 40 148 94 315 244 107 7 160 46 61 167 251 113 0 189 200 98 162 93 49 304 43 8 288 230 119 248 334 173 45 0 105 164 314 268 246 103 299 9 121 352 139 204 294 273 93 33 0 171 248 41 86 114 299 10 314 338 347 179 136 39 182 286 297 0 282 241 155 234 179 11 58 341 67 89 120 202 40 219 115 169 0 314 348 139 161 12 73 114 125 190 187 163 69 206 131 126 317 0 274 77 94 13 61 139 151 297 192 233 36 170 315 140 288 195 0 275 132 14 256 210 134 336 46 54 206 280 279 328 225 257 298 0 184 15 253 164 333 340 332 107 57 285 138 81 111 182 91 210 0

199

Table A-8: County – County Passenger Demand (Non-Business; Income Group 1).

Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 1376 1192 588 827 1428 1657 534 858 952 951 909 402 806 1621 2 791 0 1094 805 1067 932 491 664 782 1327 978 1205 1674 623 252 3 649 907 0 1392 634 475 1222 406 431 1125 1523 1435 420 1406 888 4 631 1601 1517 0 1082 1231 963 1309 540 733 1275 231 662 441 696 5 1046 1324 785 1088 0 287 395 1668 729 1149 300 1340 831 636 1605 6 683 418 911 760 323 0 981 1306 1352 192 717 456 1526 1180 519 7 774 223 297 809 1218 549 0 914 970 477 786 452 236 1471 210 8 1393 1116 579 1200 1619 839 216 0 507 796 1520 1298 1194 501 1449 9 587 1704 674 989 1425 1321 449 158 0 829 1200 199 418 554 1446 10 1520 1637 1683 868 660 188 880 1387 1437 0 1367 1167 751 1135 868 11 280 1652 324 430 580 977 192 1060 557 819 0 1521 1687 671 779 12 355 550 607 922 904 789 332 997 634 611 1536 0 1329 373 455 13 294 675 729 1438 932 1131 175 826 1524 678 1395 943 0 1330 641 14 1238 1019 647 1629 223 263 997 1354 1350 1587 1091 1243 1445 0 892 15 1226 796 1612 1646 1608 519 275 1379 669 392 538 881 440 1016 0

200


Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 2183 1891 933 1313 2267 2629 847 1361 1511 1510 1443 637 1278 2572 2 1255 0 1736 1277 1693 1479 780 1054 1241 2106 1553 1913 2657 989 400 3 1030 1439 0 2210 1006 753 1939 644 683 1785 2416 2278 666 2232 1409 4 1001 2540 2408 0 1717 1953 1528 2077 856 1164 2024 367 1051 699 1105 5 1660 2102 1245 1727 0 456 628 2647 1158 1823 476 2126 1318 1009 2548 6 1085 663 1446 1206 513 0 1558 2073 2146 305 1138 724 2421 1872 824 7 1229 354 471 1284 1933 871 0 1450 1540 756 1248 718 374 2335 333 8 2210 1771 919 1905 2569 1332 343 0 804 1264 2412 2060 1894 795 2299 9 932 2705 1070 1569 2261 2097 713 251 0 1316 1905 315 664 879 2295 10 2413 2598 2671 1378 1047 299 1397 2201 2281 0 2169 1852 1192 1801 1378 11 445 2621 514 682 921 1551 304 1683 884 1299 0 2414 2678 1065 1237 12 564 874 963 1463 1434 1253 527 1582 1006 970 2438 0 2109 592 722 13 466 1072 1157 2283 1479 1794 277 1311 2418 1076 2214 1496 0 2111 1017 14 1965 1617 1026 2586 354 417 1583 2149 2142 2519 1732 1973 2293 0 1415 15 1945 1264 2559 2612 2553 823 437 2188 1062 623 854 1398 698 1613 0

201


Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 1882 1630 804 1131 1953 2266 730 1173 1302 1301 1244 549 1102 2217 2 1082 0 1496 1100 1459 1275 672 908 1069 1815 1338 1649 2290 853 345 3 888 1240 0 1904 867 649 1671 555 589 1538 2082 1963 574 1924 1215 4 863 2189 2075 0 1480 1683 1317 1790 738 1003 1745 316 906 603 952 5 1431 1812 1073 1489 0 393 541 2281 998 1571 410 1832 1136 869 2196 6 935 572 1247 1039 442 0 1342 1786 1849 263 980 624 2087 1614 710 7 1059 305 406 1106 1666 750 0 1250 1327 652 1075 619 323 2012 287 8 1905 1527 792 1641 2214 1148 296 0 693 1089 2078 1775 1632 685 1981 9 803 2331 922 1353 1949 1807 614 216 0 1134 1642 272 572 758 1978 10 2079 2239 2301 1188 903 258 1204 1897 1966 0 1869 1596 1028 1552 1188 11 383 2259 443 587 793 1337 262 1450 762 1120 0 2081 2308 918 1066 12 486 753 830 1261 1236 1080 454 1363 867 836 2101 0 1818 510 622 13 402 924 997 1967 1274 1546 239 1130 2084 928 1908 1289 0 1819 877 14 1693 1393 884 2228 305 360 1364 1852 1846 2171 1492 1700 1976 0 1220 15 1676 1089 2205 2251 2200 710 377 1886 915 537 736 1205 602 1390 0

202


Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 808 700 345 486 838 973 313 504 559 559 534 236 473 952 2 464 0 642 472 626 547 288 390 459 779 574 708 983 366 148 3 381 532 0 817 372 279 717 238 253 660 894 843 246 826 521 4 370 940 891 0 635 722 565 768 317 430 749 136 389 259 409 5 614 778 461 639 0 169 232 979 428 674 176 786 488 373 943 6 401 245 535 446 190 0 576 767 794 113 421 268 896 693 305 7 455 131 174 475 715 322 0 536 570 280 462 266 139 864 123 8 818 655 340 705 950 493 127 0 297 467 892 762 701 294 850 9 345 1001 396 581 837 776 264 93 0 487 705 117 246 325 849 10 892 961 988 510 387 111 517 814 844 0 802 685 441 666 510 11 164 970 190 252 341 574 112 623 327 481 0 893 991 394 457 12 209 323 356 541 531 463 195 585 372 359 902 0 780 219 267 13 172 397 428 844 547 664 103 485 895 398 819 553 0 781 376 14 727 598 380 957 131 154 586 795 792 932 641 730 848 0 524 15 720 468 947 966 944 305 162 809 393 230 316 517 258 597 0

203


Orig.\ Dest. (County) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 0 559 484 239 336 580 673 217 349 387 387 370 163 327 659 2 321 0 445 327 434 379 200 270 318 539 398 490 680 253 102 3 264 369 0 566 258 193 497 165 175 457 619 583 171 572 361 4 256 651 617 0 440 500 391 532 219 298 518 94 269 179 283 5 425 538 319 442 0 117 161 678 296 467 122 544 338 258 653 6 278 170 370 309 131 0 399 531 549 78 291 186 620 479 211 7 315 91 121 329 495 223 0 371 394 194 320 184 96 598 85 8 566 454 235 488 658 341 88 0 206 324 618 527 485 204 589 9 239 693 274 402 579 537 183 64 0 337 488 81 170 225 588 10 618 665 684 353 268 77 358 564 584 0 555 474 305 461 353 11 114 671 132 175 236 397 78 431 226 333 0 618 686 273 317 12 144 224 247 375 367 321 135 405 258 248 624 0 540 152 185 13 119 275 296 585 379 460 71 336 619 276 567 383 0 541 261 14 503 414 263 662 91 107 405 550 549 645 444 505 587 0 362 15 498 324 655 669 654 211 112 560 272 159 219 358 179 413 0

204

Table A-13: Time-based Landing Fee.

Beginning of Hour

LandingFee / Operation

0030 275 0130 275 0230 275 0330 275 0430 275 0530 275 0630 600 0730 800 0830 800 0930 800 1030 600 1130 600 1230 800 1330 800 1430 600 1530 600 1630 800 1730 800 1830 800 1930 800 2030 600 2130 600 2230 275 2330 275

205

A.5 Other Predefined Input Variables

% Air Passenger Departure Time Distribution Time_Span = 15; % Minutes Slots_Total = 24 * 60 / Time_Span; Slots_HighPeak = [6:20, 34:48]; Slots_MidPeak = [1:5, 21:33, 49:56, 93:96]; Slots_LowPeak = 57:92; Weight_Slot(1) = 0.5; Weight_Slot(2) = 0.49; Weight_Slot(3) = 1 - sum(Weight_Slot(1:2)); Variation = [0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4]; % Automobile Travel Time Travel Cost Variables FirstClassTicketRatio = [0.10 0.20 0.35 0.45 0.60; ... 0.02 0.05 0.10 0.15 0.25]; AvgSpeed_Auto = 65; MaxDriveTimePerDay_Auto = [8 10]; LodgingCostPerDay_Auto = [70 80 90 100 120; ... 50 60 70 80 90]; CostPerMile_Auto = 0.15; ProcessTime_CA = [2.00 1.50 1.25 1.00; ... 0.75 0.75 0.50 0.50]; Limits of Seats Capacity = [20 400]; Limits of Coach Airfare = [5 500]; Itinerary Choice Model Utility Function Coefficients = [-0.04 -0.03 -0.01 -0.02] % 1-Airfare % 2-Duration % 3-Schedule Delay % 4-No of StopOvers

206

Appendix B: International Enplanements within the Continental U.S. (CONUS)

207

B.1 Countries Covered by each World Region

U.S. WAC Description WAC Description

1 Alaska 51 Alabama 2 Hawaii 52 Kentucky 3 Puerto Rico 53 Mississippi 4 U.S. Virgin Islands 54 Tennessee

5 U.S. Pacific Trust Territories And Possessions 61 Iowa

11 Connecticut 62 Kansas 12 Maine 63 Minnesota 13 Massachusetts 64 Missouri 14 New Hampshire 65 Nebraska 15 Rhode Island 66 North Dakota 16 Vermont 67 South Dakota 21 New Jersey 71 Arkansas 22 New York 72 Louisiana 23 Pennsylvania 73 Oklahoma 31 Delaware 74 Texas 32 District of Columbia 81 Arizona 33 Florida 82 Colorado 34 Georgia 83 Idaho 35 Maryland 84 Montana 36 North Carolina 85 Nevada 37 South Carolina 86 New Mexico 38 Virginia 87 Utah 39 West Virginia 88 Wyoming 41 Illinois 91 California 42 Indiana 92 Oregon 43 Michigan 93 Washington 44 Ohio 45 Wisconsin

Mexico

WAC Description 148 Mexico

Central America & Caribbean

WAC Description WAC Description 106 Belize 233 Cayman Islands 110 Costa Rica 235 Guadeloupe-France 118 El Salvador 238 Haiti 127 Guatemala 243 Jamaica 131 Honduras 252 Martinique-France 153 Nicaragua 256 Montserrat

208

Central America & Caribbean (Continued)WAC Description WAC Description

160 Panama Canal Zone 259 Netherlands Antilles 162 Panama Republic 273 Grenada and South Grenadines 202 Anguilla 275 St. Kitts and Nevis 204 Bahamas 276 St. Lucia 205 Barbados 277 Aruba 206 Antigua and Barbuda 279 St. Vincent and North Grenadines 207 Bermuda-UK 280 Trinidad and Tobago 219 Cuba 281 Turks and Caicos Islands-UK 221 Dominica 282 British Virgin Islands-UK 224 Dominican Republic

South America

WAC Description WAC Description 303 Argentina 344 French Guiana-France 312 Bolivia 350 Guyana 316 Brazil 365 Paraguay 324 Chile 368 Peru 327 Colombia 379 Surinam 337 Ecuador 385 Uruguay 340 Falkland Islands-UK 388 Venezuela

Europe

WAC Description WAC Description 401 Albania 451 Latvia 403 Austria 452 Lithuania 407 Azerbaijan 454 Luxembourg 409 Belgium 455 Macedonia 410 Bosnia and Herzegovina 456 Malta 411 Bulgaria 461 Netherlands 413 Belarus 465 Norway 415 Croatia 467 Poland 417 Czechoslovakia 469 Portugal 418 Czech Republic 473 Romania 419 Denmark 475 Russia (European) 422 Estonia 770 Russia (Asian) 425 Finland 477 Serbia and Montenegro 427 France 481 Slovenia 429 Germany 482 Spain 430 Berlin 483 Slovakia 431 Gibraltar-UK 484 Sweden 432 Georgia 486 Switzerland 433 Greece 488 Ukraine 437 Hungary 489 U.S.S.R. (European) 439 Iceland 493 United Kingdom 441 Ireland 497 Yugoslavia 450 Italy

209

Africa WAC Description WAC Description

500 Algeria 543 Mali 502 Angola 548 Morocco 504 Cameroons 550 Mozambique 507 Cape Verde Islands 555 Nigeria 509 Central African Republic 562 Republic Of South Africa 510 Botswana 565 Zimbabwe 515 Congo 566 Rwanda 521 Equatorial Guinea 569 Senegal 522 Ethiopia 570 Seychelles Islands 525 Djibouti 571 Sierra Leone 526 Gabon 573 Somalia 527 The Gambia 575 Namibia 529 Ghana 580 St. Helena 531 Guinea 582 Swaziland 533 Cote d Ivoire (formerly Ivory Coast) 585 Tanzania 535 Kenya 588 Tunisia 537 Liberia 590 Uganda 538 Libya 591 Arab Republic Of Egypt 541 Madagascar 597 Zambia

Middle East

WAC Description WAC Description 605 Bahrain Island 658 Oman 611 Cyprus 664 Qatar 632 Iran 667 People's Democratic Republic Of Yemen 634 Iraq 670 Saudi Arabia 636 Israel 676 Syrian Arab Republic 639 Jordan 678 United Arab Emirates 644 Kuwait 679 Turkey 647 Lebanon 694 Yemen

Asia

WAC Description WAC Description 701 Afghanistan 757 North Korea 703 Bangladesh 764 Pakistan 704 Brunei 766 Philippines 706 Myanmar 776 Singapore 707 British Indian Ocean Territory-UK 778 South Korea 709 Democratic Kampuchea (Cambodia) 781 Taiwan 713 China 782 Thailand 729 Hong Kong-China 785 Turkmenistan 733 India 786 U.S.S.R. (Asian) 736 Japan

210

Asia (Continued) WAC Description WAC Description

744 Laos 788 Uzbekistan 747 Macau 791 Vietnam 749 Malaysia

Oceania

WAC Description WAC Description 802 Australia 844 Marshall Islands 804 Papua New Guinea 845 Nauru 810 Micronesia 846 New Caledonia - France 812 Cocos Islands-Australia 851 New Zealand 813 Cook Islands-New Zealand 852 Niue-New Zealand 821 Fiji Islands 874 Solomon Islands 823 French Polynesia 881 Tonga 824 Kiribati (Gilbert and Canton Islands) 892 Western Samoa 832 Indonesia

211

B.2 Historical (1990 – 2008) and Forecast (2009 – 2040) Enplanements within the CONUS to Each of the

Nine International Regions

Figure B-1: Historical (1990 – 2008) and Forecast (2009 – 2040) Enplanements within the CONUS to Africa

(Historical Source: 1990 - 2008 T100 International Market Data).

Figure B-2: Historical (1990 – 2008) & Forecast (2009 – 2040) Enplanements within the CONUS to Asia


1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

1

2

3

4

5

6

7

8x 10

6

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

1-Africa (2008 Market Share: 0.67244%) Passengers from the CONUS 1990 - 2040

Forecast - Linear GDPUS*Africa (R2 = 0.94506; Avg. Growth Factor = 8.3691%)


1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

2-Asia (2008 Market Share: 10.197%) Passengers from the CONUS 1990 - 2040

Forecast - Linear GDPUS+Asia (R2 = 0.95567; Avg. Growth Factor = 5.4961%)


212

Figure B-3: Historical (1990 – 2008) & Forecast (2009 – 2040) Enplanements within the CONUS to Canada


Figure B-4: Historical (1990 – 2008) & Forecast (2009 – 2040) Enplanements within the CONUS to Caribbean

& Central America (Historical Source: 1990 - 2008 T100 International Market Data).

1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

0.5

1

1.5

2

2.5

3

3.5x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

3-Canada (2008 Market Share: 12.2465%) Passengers from the CONUS 1990 - 2040

Forecast - Linear GDPUS+Canada (R2 = 0.97787; Avg. Growth Factor = 3.5351%)


1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

1

2

3

4

5

6

7x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

4-Caribbean & Central America (2008 Market Share: 13.1%)

Forecast - Linear GDPUS*Caribbean & Central America (R2 = 0.98428; Avg. Growth Factor = 5.753%)


213

Figure B-5: Historical (1990 – 2008) & Forecast (2009 – 2040) Enplanements within the CONUS to Europe


Figure B-6: Historical (1990 – 2008) & Forecast (2009 – 2040) Enplanements within the CONUS to Mexico


1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20401

2

3

4

5

6

7

8x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US




1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

1

2

3

4

5

6x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

6-Mexico (2008 Market Share: 11.1382%) Passengers from the CONUS 1990 - 2040

Forecast - Linear GDPUS*Mexico (R2 = 0.95189; Avg. Growth Factor = 5.7219%)


214

Figure B-7: Historical (1990 – 2008) & Forecast (2009 – 2040) Enplanements within the CONUS to Middle

East (Historical Source: 1990 - 2008 T100 International Market Data).

Figure B-8: Historical (1990 – 2008) & Forecast (2009 – 2040) Enplanements within the CONUS to Oceania


1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

7-Middle East (2008 Market Share: 1.8976%) Passengers from the CONUS

Forecast - Linear GDPUS*Middle East (R2 = 0.92701; Avg. Growth Factor = 8.0365%)


1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400.5

1

1.5

2

2.5

3

3.5

4

4.5x 10

6

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

8-Oceania (2008 Market Share: 1.484%) Passengers from the CONUS 1990 - 2040

Forecast - Linear GDPUS+Oceania (R2 = 0.96731; Avg. Growth Factor = 3.821%)


215

Figure B-9: Historical (1990 – 2008) & Forecast (2009 – 2040) Enplanements within the CONUS to South

America (Historical Source: 1990 - 2008 T100 International Market Data).

1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

9-South America (2008 Market Share: 5.7%) Passengers from the CONUS

Forecast - Linear GDPUS+South America (R2 = 0.93717; Avg. Growth Factor = 4.3316%)


216

Table B-1: 2010 Passenger Traffic for 871 Transatlantic Airport Pairs without Nonstop Service.

1 2 3 4 5 6 7 8 9

Africa Asia Canada Caribbean & Central America Europe Mexico Middle East Oceania South

America2009 564,092 8,573,696 9,659,160 10,981,275 23,709,245 8,887,944 1,694,465 1,237,526 4,484,838 2010 609,303 9,059,093 9,961,000 11,357,571 24,419,045 9,291,731 1,845,503 1,290,894 4,777,537 2011 678,361 9,703,906 10,695,292 11,994,605 25,790,245 9,862,588 2,027,902 1,397,860 5,254,155 2012 758,750 10,366,023 11,484,587 12,759,329 27,225,171 10,435,266 2,215,403 1,507,715 5,689,992 2013 846,940 11,041,459 12,190,449 13,531,013 28,718,706 11,054,207 2,416,849 1,615,616 6,115,896 2014 940,695 11,741,906 12,891,750 14,369,569 30,276,786 11,713,542 2,630,564 1,722,404 6,563,614 2015 1,040,259 12,468,993 13,502,382 15,257,529 31,838,478 12,415,613 2,857,681 1,835,472 7,037,355 2016 1,145,816 13,222,681 14,138,215 16,215,962 33,403,020 13,163,389 3,100,624 1,954,958 7,532,981 2017 1,258,298 14,012,333 14,800,684 17,212,113 34,998,215 13,959,734 3,363,129 2,081,332 8,056,988 2018 1,378,487 14,838,688 15,491,349 18,263,343 36,622,617 14,807,823 3,644,134 2,215,014 8,607,827 2019 1,507,154 15,703,815 16,211,866 19,377,487 38,282,962 15,711,014 3,944,927 2,356,363 9,189,918 2020 1,644,596 16,609,469 16,963,975 20,554,333 39,969,878 16,672,985 4,265,929 2,505,841 9,802,207 2021 1,791,878 17,560,247 17,749,846 21,797,284 41,696,954 17,697,494 4,612,371 2,663,955 10,447,079 2022 1,949,202 18,559,053 18,571,760 23,113,850 43,447,478 18,788,511 4,984,644 2,831,200 11,130,032 2023 2,117,745 19,609,540 19,431,849 24,510,804 45,237,167 19,950,525 5,384,735 3,008,070 11,850,071 2024 2,297,961 20,715,188 20,332,327 25,992,657 47,071,682 21,188,002 5,814,679 3,195,183 12,609,533 2025 2,490,011 21,879,223 21,275,676 27,566,470 48,948,309 22,505,923 6,276,770 3,393,063 13,413,072 2026 2,693,844 23,104,308 22,264,274 29,238,886 50,871,187 23,909,615 6,773,396 3,602,368 14,262,916 2027 2,913,986 24,407,531 23,326,373 31,041,406 52,882,722 25,428,213 7,314,640 3,827,677 15,177,030 2028 3,148,231 25,780,971 24,440,270 32,956,353 54,947,654 27,047,220 7,896,917 4,066,207 16,145,946 2029 3,398,385 27,229,059 25,608,432 34,993,258 57,061,805 28,773,051 8,523,401 4,318,762 17,170,453 2030 3,663,742 28,756,776 26,833,646 37,155,802 59,230,252 30,613,021 9,197,447 4,586,107 18,257,414 2031 3,948,082 30,359,911 28,110,437 39,457,151 61,452,602 32,574,334 9,922,527 4,869,189 19,401,463 2032 4,252,767 32,042,409 29,440,961 41,905,750 63,730,146 34,665,146 10,702,496 5,168,921 20,605,552 2033 4,579,232 33,808,457 30,827,491 44,511,728 66,064,348 36,894,137 11,541,542 5,486,331 21,872,890 2034 4,929,054 35,662,446 32,272,374 47,284,584 68,456,593 39,270,273 12,444,110 5,822,404 23,206,768 2035 5,303,894 37,609,031 33,778,073 50,234,963 70,908,385 41,803,248 13,415,010 6,178,276 24,610,659 2036 5,705,535 39,653,110 35,347,147 53,374,794 73,421,247 44,503,563 14,459,452 6,555,117 26,088,287 2037 6,135,911 41,799,839 36,982,262 56,716,034 75,996,701 47,382,315 15,582,968 6,954,123 27,643,515 2038 6,597,059 44,054,662 38,686,197 60,271,308 78,636,349 50,451,028 16,791,529 7,376,621 29,280,370 2039 7,091,197 46,423,340 40,461,850 64,054,701 81,341,850 53,722,387 18,091,647 7,823,982 31,003,200 2040 7,620,671 48,911,910 42,312,234 68,081,050 84,114,806 57,209,762 19,490,171 8,297,680 32,816,460

Year\Region

217

B.3 Comparison between Historical and Forecast Enplanements within the CONUS to Each of the Nine

International Regions during 1990 – 2008


Africa during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).


Asia during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20081

1.5

2

2.5

3

3.5

4

4.5

5

5.5

6x 10

5

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

1-Africa (2008 Market Share: 0.67244%) Passengers from the CONUS 1990 - 2008

Forecast - Linear GDPUS*Africa (R2 = 0.94506)

Historical

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20083

4

5

6

7

8

9

10x 10

6

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

2-Asia (2008 Market Share: 10.197%) Passengers from the CONUS 1990 - 2008

Forecast - Linear GDPUS+Asia (R2 = 0.95567)

Historical

218


Canada during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).


Caribbean & Central America during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20084

5

6

7

8

9

10

11x 10

6

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

3-Canada (2008 Market Share: 12.2465%) Passengers from the CONUS 1990 - 2008

Forecast - Linear GDPUS+Canada (R2 = 0.97787)

Historical

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20084

5

6

7

8

9

10

11

12x 10

6

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

4-Caribbean & Central America (2008 Market Share: 13.1%) Passengers

Forecast - Linear GDPUS*Caribbean & Central America (R2 = 0.98428)

Historical

219


Europe during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).


Mexico during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20081

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US



Historical

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20083

4

5

6

7

8

9

10x 10

6

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

6-Mexico (2008 Market Share: 11.1382%) Passengers from the CONUS 1990 - 2008

Forecast - Linear GDPUS*Mexico (R2 = 0.95189)

Historical

220


Middle East during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).


Oceania during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20080.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8x 10

6

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

7-Middle East (2008 Market Share: 1.9%) Passengers from the CONUS 1990 - 2008

Forecast - Linear GDPUS*Middle East (R2 = 0.92701)

Historical

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20085

6

7

8

9

10

11

12

13

14x 10

5

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

8-Oceania (2008 Market Share: 1.484%) Passengers from the CONUS 1990 - 2008

Forecast - Linear GDPUS+Oceania (R2 = 0.96731)

Historical

221


South America during 1990 – 2008 (Historical Source: 1990 - 2008 T100 International Market Data).

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20081.5

2

2.5

3

3.5

4

4.5

5x 10

6

Year

MA

RK

ET P

asse

nger

s fr

om th

e C

ON

US

9-South America (2008 Market Share: 5.7%) Passengers from the CONUS

Forecast - Linear GDPUS+South America (R2 = 0.93717)

Historical

222

B.4 Input Data for the Model

Table B-2: Enplanements from the CONUS to the Nine International Regions during 1990 – 2008 (Source: 1990 – 2008 T100

International Market Data).

Year\Region 1 2 3 4 5 6 7 8 9

Sum Africa Asia Canada

Caribbean & Central America

Europe Mexico Middle East Oceania South

America

1990 109,132 3,783,816 5,333,482 4,683,693 12,470,610 4,169,332 371,556 500,628 1,868,447 33,290,6961991 103,674 3,932,105 4,769,731 4,829,814 11,584,914 4,350,890 376,579 543,279 2,003,672 32,494,6581992 129,816 4,382,668 4,958,315 4,655,113 13,564,309 4,432,115 464,244 605,039 2,274,517 35,466,1361993 146,144 4,719,005 5,253,884 5,223,746 14,209,112 4,448,911 517,258 702,374 2,589,315 37,809,7491994 160,003 5,176,847 5,182,761 5,290,952 15,017,167 4,633,636 568,807 782,361 2,905,731 39,718,2651995 203,031 5,796,956 5,929,238 5,775,749 15,871,988 4,510,954 656,488 842,064 3,353,400 42,939,8681996 226,718 6,321,883 6,978,233 5,777,917 16,933,222 5,220,201 720,487 914,538 3,471,606 46,564,8051997 277,310 6,766,916 7,272,258 5,891,102 18,590,263 5,623,287 754,556 932,224 3,930,021 50,037,9371998 283,065 6,496,658 7,788,396 6,622,929 20,424,287 5,793,809 769,908 927,106 4,141,494 53,247,6521999 332,288 7,092,700 7,896,911 7,201,638 22,163,883 6,152,538 802,175 1,017,298 4,034,360 56,693,7912000 352,194 7,786,632 8,510,877 7,513,390 24,154,020 6,746,507 866,714 1,167,153 4,120,668 61,218,1552001 337,771 7,078,230 8,032,883 7,483,172 21,537,579 6,409,327 698,361 1,059,245 3,890,136 56,526,7042002 274,727 7,197,647 7,991,401 7,614,264 20,368,716 6,159,220 642,782 1,037,040 3,470,151 54,755,9482003 291,296 6,372,177 8,099,678 8,222,714 20,277,036 6,409,356 649,790 1,052,171 3,331,491 54,705,7092004 309,560 7,717,432 9,129,132 9,106,124 22,397,932 7,862,752 852,371 1,145,276 3,654,381 62,174,9602005 328,176 8,416,467 9,702,576 9,818,181 23,511,515 8,931,096 954,675 1,198,364 4,035,643 66,896,6932006 370,257 8,783,470 10,101,535 10,565,388 24,238,097 9,295,596 1,117,521 1,270,134 4,257,897 69,999,8952007 483,171 9,025,758 10,459,439 11,130,631 25,419,502 9,620,642 1,389,851 1,300,756 4,641,757 73,471,507

2008 582,136 8,827,612 10,601,871 11,367,361 25,817,774 9,642,418 1,642,726 1,284,697 4,962,426 74,729,021

223

Table B-3: Growth of Enplanements from the CONUS to the Nine International Regions during 1990 – 2008 (Source: 1990 – 2008

T100 International Market Data).

Year\Region

1 2 3 4 5 6 7 8 9

Sum Africa Asia Canada



America

1990 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 1991 ‐5.0% 3.9% ‐10.6% 3.1% ‐7.1% 4.4% 1.4% 8.5% 7.2% ‐2.4% 1992 25.2% 11.5% 4.0% ‐3.6% 17.1% 1.9% 23.3% 11.4% 13.5% 9.1% 1993 12.6% 7.7% 6.0% 12.2% 4.8% 0.4% 11.4% 16.1% 13.8% 6.6% 1994 9.5% 9.7% ‐1.4% 1.3% 5.7% 4.2% 10.0% 11.4% 12.2% 5.0% 1995 26.9% 12.0% 14.4% 9.2% 5.7% ‐2.6% 15.4% 7.6% 15.4% 8.1% 1996 11.7% 9.1% 17.7% 0.0% 6.7% 15.7% 9.7% 8.6% 3.5% 8.4% 1997 22.3% 7.0% 4.2% 2.0% 9.8% 7.7% 4.7% 1.9% 13.2% 7.5% 1998 2.1% ‐4.0% 7.1% 12.4% 9.9% 3.0% 2.0% ‐0.5% 5.4% 6.4% 1999 17.4% 9.2% 1.4% 8.7% 8.5% 6.2% 4.2% 9.7% ‐2.6% 6.5% 2000 6.0% 9.8% 7.8% 4.3% 9.0% 9.7% 8.0% 14.7% 2.1% 8.0% 2001 ‐4.1% ‐9.1% ‐5.6% ‐0.4% ‐10.8% ‐5.0% ‐19.4% ‐9.2% ‐5.6% ‐7.7% 2002 ‐18.7% 1.7% ‐0.5% 1.8% ‐5.4% ‐3.9% ‐8.0% ‐2.1% ‐10.8% ‐3.1% 2003 6.0% ‐11.5% 1.4% 8.0% ‐0.5% 4.1% 1.1% 1.5% ‐4.0% ‐0.1% 2004 6.3% 21.1% 12.7% 10.7% 10.5% 22.7% 31.2% 8.8% 9.7% 13.7%2005 6.0% 9.1% 6.3% 7.8% 5.0% 13.6% 12.0% 4.6% 10.4% 7.6% 2006 12.8% 4.4% 4.1% 7.6% 3.1% 4.1% 17.1% 6.0% 5.5% 4.6% 2007 30.5% 2.8% 3.5% 5.3% 4.9% 3.5% 24.4% 2.4% 9.0% 5.0% 2008 20.5% ‐2.2% 1.4% 2.1% 1.6% 0.2% 18.2% ‐1.2% 6.9% 1.7%

Average Annual Growth 1990‐2008 9.7% 4.8% 3.9% 5.0% 4.1% 4.8% 8.6% 5.4% 5.6% 4.6%

224

Table B-4: Real GDP of U.S. and the Nine International Regions during 1990 – 2008: Billion 2005 $ (Source: United States

Department of Agriculture (USDA) International Macroeconomic Data Set).

Year\Region

1 2 3 4 5 6 7 8 9

U.S. Africa Asia Canada



America

1990 674.32 5,703.81 798.14 184.07 11,783.31 499.65 717.62 502.69 1,094.70 8,037.13 1991 675.09 5,958.98 781.43 184.15 11,775.60 520.65 761.51 503.80 1,131.67 8,023.57 1992 670.07 6,173.57 788.27 186.41 11,710.59 539.08 792.44 520.76 1,157.93 8,290.36 1993 662.12 6,375.85 806.71 188.22 11,606.44 549.52 833.69 542.89 1,212.63 8,511.95 1994 671.63 6,629.76 845.48 193.84 11,813.59 574.03 839.64 566.28 1,276.43 8,854.12 1995 695.72 6,944.54 869.21 201.99 12,091.87 538.32 873.19 589.71 1,327.93 9,075.82 1996 729.94 7,292.16 883.28 209.17 12,289.34 566.06 916.30 611.76 1,363.25 9,411.66 1997 748.63 7,558.50 920.60 218.42 12,619.08 604.40 961.91 635.98 1,425.96 9,834.96 1998 773.33 7,503.22 958.33 228.46 12,947.81 634.06 1,025.97 665.21 1,438.19 10,245.60 1999 789.31 7,740.81 1,011.35 239.94 13,327.71 658.53 1,055.52 691.48 1,419.51 10,701.44 2000 817.13 8,125.22 1,064.58 248.71 13,885.56 702.03 1,101.81 705.73 1,463.68 11,093.21 2001 844.68 8,322.81 1,085.02 255.97 14,181.64 700.87 1,080.01 732.52 1,478.47 11,176.49 2002 875.06 8,613.00 1,120.61 261.96 14,381.32 706.70 1,097.48 757.20 1,480.46 11,355.14 2003 916.22 8,989.34 1,139.77 269.56 14,604.13 716.26 1,093.09 785.54 1,506.37 11,640.13 2004 961.62 9,478.42 1,174.95 278.18 14,990.42 746.25 1,201.34 810.66 1,609.14 12,063.88 2005 1,013.59 9,966.09 1,207.10 286.08 15,313.20 767.17 1,268.59 830.70 1,685.85 12,433.39 2006 1,071.65 10,554.11 1,240.56 299.25 15,832.28 803.77 1,340.40 853.57 1,774.20 12,790.92 2007 1,136.83 11,177.96 1,274.22 312.09 16,346.09 830.25 1,406.15 888.36 1,885.14 13,064.85 2008 1,200.14 11,556.76 1,279.31 319.64 16,529.14 841.87 1,487.79 905.04 1,985.27 13,122.12

225

Table B-5: Growth of Real GDP of the U.S. and the Nine International Regions during 1990 – 2008 (Source: United States


Year\Region

1 2 3 4 5 6 7 8 9




America

1990 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 1991 0.1% 4.5% ‐2.1% 0.0% ‐0.1% 4.2% 6.1% 0.2% 3.4% ‐0.2% 1992 ‐0.7% 3.6% 0.9% 1.2% ‐0.6% 3.5% 4.1% 3.4% 2.3% 3.3% 1993 ‐1.2% 3.3% 2.3% 1.0% ‐0.9% 1.9% 5.2% 4.2% 4.7% 2.7% 1994 1.4% 4.0% 4.8% 3.0% 1.8% 4.5% 0.7% 4.3% 5.3% 4.0% 1995 3.6% 4.7% 2.8% 4.2% 2.4% ‐6.2% 4.0% 4.1% 4.0% 2.5% 1996 4.9% 5.0% 1.6% 3.6% 1.6% 5.2% 4.9% 3.7% 2.7% 3.7% 1997 2.6% 3.7% 4.2% 4.4% 2.7% 6.8% 5.0% 4.0% 4.6% 4.5% 1998 3.3% ‐0.7% 4.1% 4.6% 2.6% 4.9% 6.7% 4.6% 0.9% 4.2% 1999 2.1% 3.2% 5.5% 5.0% 2.9% 3.9% 2.9% 3.9% ‐1.3% 4.4% 2000 3.5% 5.0% 5.3% 3.7% 4.2% 6.6% 4.4% 2.1% 3.1% 3.7% 2001 3.4% 2.4% 1.9% 2.9% 2.1% ‐0.2% ‐2.0% 3.8% 1.0% 0.8% 2002 3.6% 3.5% 3.3% 2.3% 1.4% 0.8% 1.6% 3.4% 0.1% 1.6% 2003 4.7% 4.4% 1.7% 2.9% 1.5% 1.4% ‐0.4% 3.7% 1.8% 2.5% 2004 5.0% 5.4% 3.1% 3.2% 2.6% 4.2% 9.9% 3.2% 6.8% 3.6% 2005 5.4% 5.1% 2.7% 2.8% 2.2% 2.8% 5.6% 2.5% 4.8% 3.1% 2006 5.7% 5.9% 2.8% 4.6% 3.4% 4.8% 5.7% 2.8% 5.2% 2.9% 2007 6.1% 5.9% 2.7% 4.3% 3.2% 3.3% 4.9% 4.1% 6.3% 2.1% 2008 12.0% 9.5% 3.1% 6.8% 4.4% 4.7% 11.0% 6.0% 11.9% 2.6%

Average Annual Growth 1990‐2008

3.3% 4.0% 2.7% 3.1% 1.9% 2.9% 4.1% 3.3% 3.4% 2.8%

226

Table B-6: Real Per Capita GDP of the U.S. and the Nine International Regions during 1990 – 2008: 2005 $ (Source: United States


Year\Region 1 2 3 4 5 6 7 8 9




America

1990 1,069 1,919 28,720 3,030 15,933 5,884 3,960 18,283 3,688 32,132 1991 1,042 1,973 27,791 2,973 15,866 6,020 4,064 18,011 3,748 31,652 1992 1,008 2,012 27,615 2,959 15,737 6,118 4,093 18,323 3,770 32,271 1993 972 2,046 27,862 2,938 15,573 6,123 4,172 18,818 3,882 32,706 1994 964 2,096 28,826 2,972 15,832 6,285 4,036 19,357 4,018 33,610 1995 975 2,163 29,276 3,042 16,191 5,796 4,106 19,854 4,112 34,048 1996 999 2,239 29,417 3,096 16,450 5,997 4,283 20,270 4,155 34,901 1997 1,001 2,288 30,377 3,180 16,888 6,303 4,391 20,753 4,279 36,037 1998 1,009 2,241 31,367 3,274 17,324 6,515 4,478 21,401 4,250 37,106 1999 1,006 2,283 32,814 3,387 17,835 6,678 4,416 21,933 4,135 38,316 2000 1,017 2,367 34,231 3,458 18,585 7,025 4,575 22,167 4,205 39,314 2001 1,026 2,395 34,580 3,506 18,984 6,922 4,485 22,640 4,190 39,210 2002 1,038 2,450 35,417 3,534 19,257 6,896 4,549 23,015 4,139 39,465 2003 1,062 2,528 35,741 3,583 19,562 6,906 4,668 23,498 4,154 40,109 2004 1,089 2,636 36,563 3,643 20,085 7,110 4,904 23,898 4,377 41,189 2005 1,122 2,742 37,272 3,692 20,522 7,224 5,108 24,145 4,523 42,067 2006 1,160 2,873 37,988 3,805 21,224 7,480 5,371 24,450 4,697 42,870 2007 1,202 3,011 38,688 3,911 21,919 7,638 5,515 25,072 4,927 43,363 2008 1,241 3,080 38,519 3,949 22,173 7,657 5,661 25,172 5,123 43,156

227

Table B-7: Growth of Real Per Capita GDP of the U.S. and the Nine International Regions during 1990 – 2008 (Source: United States


Year\Region

1 2 3 4 5 6 7 8 9

U.S. Africa Asia CanadaCaribbean & Central America


America

1990 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 1991 ‐2.5% 2.8% ‐3.2% ‐1.9% ‐0.4% 2.3% 2.6% ‐1.5% 1.6% ‐1.5%1992 ‐3.2% 2.0% ‐0.6% ‐0.5% ‐0.8% 1.6% 0.7% 1.7% 0.6% 2.0% 1993 ‐3.6% 1.7% 0.9% ‐0.7% ‐1.0% 0.1% 1.9% 2.7% 3.0% 1.3% 1994 ‐0.9% 2.4% 3.5% 1.2% 1.7% 2.6% ‐3.3% 2.9% 3.5% 2.8% 1995 1.2% 3.2% 1.6% 2.4% 2.3% ‐7.8% 1.7% 2.6% 2.3% 1.3% 1996 2.5% 3.5% 0.5% 1.8% 1.6% 3.5% 4.3% 2.1% 1.0% 2.5% 1997 0.2% 2.2% 3.3% 2.7% 2.7% 5.1% 2.5% 2.4% 3.0% 3.3% 1998 0.9% ‐2.0% 3.3% 2.9% 2.6% 3.4% 2.0% 3.1% ‐0.7% 3.0% 1999 ‐0.4% 1.9% 4.6% 3.4% 2.9% 2.5% ‐1.4% 2.5% ‐2.7% 3.3% 2000 1.1% 3.7% 4.3% 2.1% 4.2% 5.2% 3.6% 1.1% 1.7% 2.6% 2001 0.9% 1.2% 1.0% 1.4% 2.1% ‐1.5% ‐2.0% 2.1% ‐0.4% ‐0.3%2002 1.2% 2.3% 2.4% 0.8% 1.4% ‐0.4% 1.4% 1.7% ‐1.2% 0.6% 2003 2.3% 3.2% 0.9% 1.4% 1.6% 0.1% 2.6% 2.1% 0.4% 1.6% 2004 2.6% 4.3% 2.3% 1.7% 2.7% 3.0% 5.1% 1.7% 5.4% 2.7% 2005 3.0% 4.0% 1.9% 1.3% 2.2% 1.6% 4.1% 1.0% 3.3% 2.1% 2006 3.3% 4.8% 1.9% 3.1% 3.4% 3.6% 5.2% 1.3% 3.9% 1.9% 2007 3.7% 4.8% 1.8% 2.8% 3.3% 2.1% 2.7% 2.5% 4.9% 1.1% 2008 7.0% 7.2% 1.4% 3.8% 4.5% 2.4% 5.4% 3.0% 9.1% 0.7%

Average Annual Growth (1990 – 2008)

0.8% 2.7% 1.6% 1.5% 1.9% 1.5% 2.0% 1.8% 1.8% 1.7%

228

Table B-8: Real GDP of U.S. and the Nine International Regions during 2009 – 2040: Billion 2005 $ (Source: United States


1 2 3 4 5 6 7 8 9

Africa Asia Canada Caribbean & Central America Europe Mexico Middle East Oceania South America

2009 1,206.07 11,482.58 1,247.33 315.31 15,777.20 788.84 1,565.00 899.19 1,965.91 12,794.07 2010 1,237.38 11,931.65 1,267.29 319.53 15,820.91 808.56 1,638.08 909.33 2,032.14 13,113.92 2011 1,288.89 12,533.45 1,317.98 329.16 16,103.59 836.86 1,717.47 943.42 2,127.40 13,533.57 2012 1,355.56 13,203.37 1,377.29 343.01 16,459.62 866.15 1,800.87 979.95 2,218.81 13,912.51 2013 1,428.64 13,897.69 1,428.94 356.41 16,848.97 898.19 1,889.71 1,013.70 2,308.24 14,288.15 2014 1,503.62 14,621.49 1,478.95 370.76 17,261.24 931.43 1,980.71 1,044.40 2,401.11 14,673.93 2015 1,580.46 15,377.08 1,519.62 385.47 17,664.95 965.89 2,074.21 1,076.07 2,498.25 15,070.12 2016 1,659.04 16,164.10 1,561.41 401.04 18,059.41 1,001.63 2,171.30 1,108.62 2,598.49 15,477.02 2017 1,740.07 16,997.11 1,604.35 416.51 18,458.59 1,038.69 2,273.74 1,142.13 2,703.17 15,894.90 2018 1,824.01 17,876.98 1,648.47 432.29 18,861.45 1,077.12 2,380.24 1,176.64 2,811.69 16,324.06 2019 1,911.24 18,806.68 1,693.80 448.51 19,271.13 1,116.97 2,490.97 1,212.15 2,924.84 16,764.81 2020 2,001.60 19,788.68 1,740.38 465.09 19,682.52 1,158.30 2,605.57 1,248.70 3,042.16 17,217.46 2021 2,095.69 20,829.94 1,788.24 482.04 20,102.26 1,201.16 2,726.32 1,286.34 3,163.92 17,682.33 2022 2,193.19 21,934.73 1,837.42 499.46 20,521.46 1,245.60 2,852.66 1,325.10 3,291.07 18,159.75 2023 2,294.71 23,108.50 1,887.95 517.42 20,947.83 1,291.69 2,984.88 1,365.00 3,423.03 18,650.06 2024 2,400.11 24,356.40 1,939.87 535.93 21,383.91 1,339.48 3,123.23 1,406.10 3,559.99 19,153.62 2025 2,508.98 25,683.21 1,993.21 555.05 21,827.98 1,389.04 3,268.03 1,448.41 3,702.65 19,670.76 2026 2,620.74 27,092.76 2,048.02 574.82 22,281.77 1,440.44 3,419.57 1,491.98 3,851.12 20,201.87 2027 2,736.07 28,591.54 2,104.35 595.19 22,746.42 1,493.73 3,578.17 1,536.84 4,004.99 20,767.53 2028 2,854.81 30,185.72 2,162.21 616.21 23,222.59 1,549.00 3,744.16 1,583.03 4,165.28 21,349.02 2029 2,977.82 31,881.87 2,221.68 637.95 23,707.69 1,606.31 3,917.91 1,630.60 4,331.69 21,946.79 2030 3,103.88 33,687.42 2,282.77 660.36 24,203.86 1,665.75 4,099.78 1,679.57 4,505.14 22,561.30

2031 3,235.28 35,595.23 2,345.55 683.56 24,710.43 1,727.38 4,290.09 1,730.01 4,685.55 23,193.02 2032 3,372.25 37,611.08 2,410.05 707.57 25,227.59 1,791.29 4,489.23 1,781.96 4,873.17 23,842.42 2033 3,515.01 39,741.09 2,476.33 732.43 25,755.58 1,857.57 4,697.62 1,835.48 5,068.31 24,510.01 2034 3,663.82 41,991.72 2,544.43 758.16 26,294.61 1,926.30 4,915.68 1,890.60 5,271.26 25,196.29 2035 3,818.93 44,369.82 2,614.40 784.79 26,844.93 1,997.57 5,143.86 1,947.38 5,482.34 25,901.78 2036 3,980.60 46,882.60 2,686.29 812.36 27,406.77 2,071.48 5,382.64 2,005.87 5,701.87 26,627.03 2037 4,149.12 49,537.68 2,760.17 840.90 27,980.36 2,148.13 5,632.50 2,066.11 5,930.20 27,372.59 2038 4,324.77 52,343.12 2,836.07 870.44 28,565.96 2,227.61 5,893.95 2,128.16 6,167.66 28,139.02 2039 4,507.86 55,307.45 2,914.06 901.02 29,163.82 2,310.03 6,167.55 2,192.07 6,414.64 28,926.92 2040 4,698.70 58,439.65 2,994.20 932.68 29,774.18 2,395.50 6,453.84 2,257.90 6,671.50 29,736.87

Year\Region U.S.

229

Table B-9: Growth of Real GDP of U.S. and the Nine International Regions during 2009 – 2040 (Source: United States Department of

Agriculture (USDA) International Macroeconomic Data Set).

1 2 3 4 5 6 7 8 9


2009 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐2010 2.6% 3.9% 1.6% 1.3% 0.3% 2.5% 4.7% 1.1% 3.4% 2.5%2011 4.2% 5.0% 4.0% 3.0% 1.8% 3.5% 4.8% 3.7% 4.7% 3.2%2012 5.2% 5.3% 4.5% 4.2% 2.2% 3.5% 4.9% 3.9% 4.3% 2.8%2013 5.4% 5.3% 3.8% 3.9% 2.4% 3.7% 4.9% 3.4% 4.0% 2.7%2014 5.2% 5.2% 3.5% 4.0% 2.4% 3.7% 4.8% 3.0% 4.0% 2.7%2015 5.1% 5.2% 2.8% 4.0% 2.3% 3.7% 4.7% 3.0% 4.0% 2.7%2016 5.0% 5.1% 2.8% 4.0% 2.2% 3.7% 4.7% 3.0% 4.0% 2.7%2017 4.9% 5.2% 2.8% 3.9% 2.2% 3.7% 4.7% 3.0% 4.0% 2.7%2018 4.8% 5.2% 2.8% 3.8% 2.2% 3.7% 4.7% 3.0% 4.0% 2.7%2019 4.8% 5.2% 2.8% 3.8% 2.2% 3.7% 4.7% 3.0% 4.0% 2.7%2020 4.7% 5.2% 2.8% 3.7% 2.1% 3.7% 4.6% 3.0% 4.0% 2.7%2021 4.7% 5.3% 2.8% 3.6% 2.1% 3.7% 4.6% 3.0% 4.0% 2.7%2022 4.7% 5.3% 2.8% 3.6% 2.1% 3.7% 4.6% 3.0% 4.0% 2.7%2023 4.6% 5.4% 2.8% 3.6% 2.1% 3.7% 4.6% 3.0% 4.0% 2.7%2024 4.6% 5.4% 2.8% 3.6% 2.1% 3.7% 4.6% 3.0% 4.0% 2.7%2025 4.5% 5.4% 2.8% 3.6% 2.1% 3.7% 4.6% 3.0% 4.0% 2.7%2026 4.5% 5.5% 2.8% 3.6% 2.1% 3.7% 4.6% 3.0% 4.0% 2.7%2027 4.4% 5.5% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2028 4.3% 5.6% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2029 4.3% 5.6% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2030 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%

2031 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2032 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2033 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2034 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2035 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2036 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2037 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2038 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2039 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%2040 4.2% 5.7% 2.8% 3.5% 2.1% 3.7% 4.6% 3.0% 4.0% 2.8%

U.S.Year\Region

230

Table B-10: Real Per Capita GDP of U.S. and Nine International Regions during 2009 – 2040: 2005 $ (Source: United States


1 2 3 4 5 6 7 8 9

Africa Asia CanadaCaribbean &

Central America Europe Mexico Middle East Oceania South America

2009 1,219 3,029 37,248 3,841 21,173 7,093 5,415 24,662 5,010 41,646 2010 1,223 3,115 37,538 3,838 21,242 7,189 5,497 24,609 5,116 42,271 2011 1,245 3,239 38,729 3,900 21,634 7,359 5,666 25,198 5,293 43,206 2012 1,281 3,378 40,154 4,010 22,128 7,533 5,861 25,837 5,457 43,990 2013 1,321 3,521 41,337 4,111 22,669 7,728 6,058 26,390 5,612 44,744 2014 1,360 3,668 42,456 4,221 23,245 7,930 6,256 26,851 5,774 45,511 2015 1,399 3,822 43,294 4,333 23,813 8,138 6,454 27,328 5,943 46,293 2016 1,438 3,981 44,154 4,451 24,372 8,353 6,656 27,817 6,116 47,089 2017 1,477 4,149 45,036 4,566 24,943 8,576 6,864 28,320 6,297 47,900 2018 1,517 4,326 45,942 4,681 25,523 8,806 7,079 28,838 6,485 48,728 2019 1,557 4,513 46,872 4,799 26,117 9,045 7,301 29,371 6,680 49,572 2020 1,598 4,710 47,829 4,918 26,719 9,292 7,531 29,921 6,882 50,434 2021 1,640 4,919 48,814 5,039 27,338 9,548 7,772 30,488 7,091 51,314 2022 1,682 5,141 49,829 5,163 27,962 9,813 8,023 31,072 7,309 52,213 2023 1,726 5,377 50,876 5,290 28,602 10,088 8,284 31,676 7,536 53,131 2024 1,771 5,628 51,955 5,421 29,262 10,373 8,556 32,299 7,771 54,070 2025 1,816 5,895 53,069 5,556 29,939 10,669 8,838 32,942 8,015 55,031 2026 1,861 6,179 54,219 5,695 30,636 10,975 9,132 33,606 8,270 56,013 2027 1,907 6,481 55,406 5,838 31,354 11,293 9,437 34,290 8,533 57,073 2028 1,953 6,802 56,629 5,986 32,096 11,624 9,754 34,996 8,807 58,157 2029 2,000 7,144 57,892 6,139 32,857 11,967 10,083 35,726 9,092 59,268 2030 2,047 7,508 59,193 6,297 33,641 12,323 10,424 36,481 9,390 60,404

2031 2,095 7,891 60,524 6,458 34,443 12,690 10,778 37,251 9,697 61,563 2032 2,144 8,293 61,884 6,624 35,264 13,068 11,143 38,038 10,014 62,743 2033 2,195 8,716 63,276 6,794 36,105 13,457 11,521 38,842 10,341 63,947 2034 2,246 9,160 64,698 6,969 36,966 13,858 11,911 39,662 10,680 65,173 2035 2,299 9,627 66,153 7,148 37,848 14,271 12,315 40,500 11,029 66,423 2036 2,353 10,118 67,640 7,331 38,751 14,696 12,733 41,355 11,390 67,696 2037 2,409 10,633 69,161 7,520 39,675 15,133 13,164 42,229 11,762 68,995 2038 2,465 11,175 70,715 7,713 40,621 15,584 13,611 43,121 12,147 70,318 2039 2,523 11,745 72,305 7,911 41,590 16,048 14,072 44,031 12,544 71,666 2040 2,583 12,343 73,931 8,114 42,582 16,526 14,549 44,961 12,954 73,041

U.S.Year\Region

231

Table B-11: Growth of Real Per Capita GDP of U.S. and Nine International Regions during 2009 – 2040 (Source: United States


1 2 3 4 5 6 7 8 9


2009 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐2010 0.3% 2.8% 0.8% ‐0.1% 0.3% 1.4% 1.5% ‐0.2% 2.1% 1.5%2011 1.9% 4.0% 3.2% 1.6% 1.8% 2.4% 3.1% 2.4% 3.4% 2.2%2012 2.9% 4.3% 3.7% 2.8% 2.3% 2.4% 3.4% 2.5% 3.1% 1.8%2013 3.1% 4.2% 2.9% 2.5% 2.4% 2.6% 3.4% 2.1% 2.9% 1.7%2014 3.0% 4.2% 2.7% 2.7% 2.5% 2.6% 3.3% 1.7% 2.9% 1.7%2015 2.9% 4.2% 2.0% 2.6% 2.4% 2.6% 3.2% 1.8% 2.9% 1.7%2016 2.8% 4.2% 2.0% 2.7% 2.4% 2.6% 3.1% 1.8% 2.9% 1.7%2017 2.7% 4.2% 2.0% 2.6% 2.3% 2.7% 3.1% 1.8% 3.0% 1.7%2018 2.7% 4.3% 2.0% 2.5% 2.3% 2.7% 3.1% 1.8% 3.0% 1.7%2019 2.7% 4.3% 2.0% 2.5% 2.3% 2.7% 3.1% 1.9% 3.0% 1.7%2020 2.6% 4.4% 2.0% 2.5% 2.3% 2.7% 3.1% 1.9% 3.0% 1.7%2021 2.6% 4.4% 2.1% 2.5% 2.3% 2.8% 3.2% 1.9% 3.0% 1.7%2022 2.6% 4.5% 2.1% 2.5% 2.3% 2.8% 3.2% 1.9% 3.1% 1.8%2023 2.6% 4.6% 2.1% 2.5% 2.3% 2.8% 3.3% 1.9% 3.1% 1.8%2024 2.6% 4.7% 2.1% 2.5% 2.3% 2.8% 3.3% 2.0% 3.1% 1.8%2025 2.6% 4.7% 2.1% 2.5% 2.3% 2.8% 3.3% 2.0% 3.1% 1.8%2026 2.5% 4.8% 2.2% 2.5% 2.3% 2.9% 3.3% 2.0% 3.2% 1.8%2027 2.5% 4.9% 2.2% 2.5% 2.3% 2.9% 3.3% 2.0% 3.2% 1.9%2028 2.4% 5.0% 2.2% 2.5% 2.4% 2.9% 3.4% 2.1% 3.2% 1.9%2029 2.4% 5.0% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.2% 1.9%2030 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%

2031 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%2032 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%2033 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%2034 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%2035 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%2036 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%2037 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%2038 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%2039 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%2040 2.4% 5.1% 2.2% 2.6% 2.4% 3.0% 3.4% 2.1% 3.3% 1.9%

U.S.Year\Region

232

B.5 Summary of Matlab Functions and Data Preparation

T100I_MARKET_1990_2008_121009.accdb T100D_MARKET_1990_2008_121009.accdb Input T100I_MARKET_ALL_CARRIER_1990_2008

T100D_MARKET_ALL_CARRIER_1990_2008 Output T100I_MARKET_US_Airport to WorldRegion_Historical_121009.txt

T100D_MARKET_US_Airport to AK_HI_Others_Historical_121009.txt Summarize_T100_US_Airport_to_WorldRegion_Pax_Historical.m % This function: % 1. Summarizes the passengers from each U.S. airport to each of 12 world % regions during the historical years % % 2. Summarizes the passengers from each U.S. airport to all 12 world % regions during the historical years % % 3. Summarizes the passengers from all U.S. airports to each of 12 world % regions during the historical yearsInput WorldRegion_Wac.mat (Predefined input)

US_Airport_to_WorldRegion_Historical_Raw.mat (This is combination of T100I_MARKET_US_Airport to WorldRegion_Historical_121009.txt T100D_MARKET_US_Airport to AK_HI_Others_Historical_121009.txt)

Output T100_US_to_WorldRegion_Pax_Historical.mat T100_US_Airport_to_WorldRegion_Pax_Historical.mat Col 2 - 13 'Africa' 'Asia' 'Canada' 'Caribbean & Central America' 'Europe' 'Mexico' 'Middle East' 'Oceania' 'South America' 'Alaska' 'Hawaii' 'Others' Col – 14 : Sum of 12 regions

Select_Focus_US_Airports.m % This function : % Selects the focus US airports based on the total number of % international (total 12 regions) passengers at each airport and the % predefined threshold number of international passengersInput Threshold_International_Pax (Predefined variable = 10,000)

T100_US_Airport_to_WorldRegion_Pax_Historical.matOutput Focus_US_Airports.mat

233

Filter_T100_US_Airport_to_WorldRegion_Pax.m % This function: % Filters the output 'US_Airport_to_WorldRegion_Pax' by the U.S. airport % from 'Focus_US_Airports' and combine all the other airports as 'Other'Input Focus_US_Airports.mat

T100_US_Airport_to_WorldRegion_Pax_Historical.matOutput Focus_T100_US_Airport_to_WorldRegion_Pax_Historical.mat

Col 2 - 13 'Africa' 'Asia' 'Canada' 'Caribbean & Central America' 'Europe' 'Mexico' 'Middle East' 'Oceania' 'South America' 'Alaska' 'Hawaii' 'Others' Col – 14 : Sum of 12 regions

Calibrate_Model_Coefficient.m % This function calibrates the coefficients for each of 8 candidate models % Model_Index = 1 (Linear - SGDP); % Model_Index = 2 (Linear - PGDP); % Model_Index = 3 (Linear - S-PerCapitaIncome); % Model_Index = 4 (Linear - P-PerCapitaIncome); % Model_Index = 5 (Log Linear - SGDP); % Model_Index = 6 (Log Linear - PGDP); % Model_Index = 7 (Log Linear - S-PerCapitaIncome); % Model_Index = 8 (Log Linear - P-PerCapitaIncome); % Model_Index = 9 (Selected - Best of 1-4);Input GDP_2005_Dollar_Analyze_Years.mat (Predefined input)

PerCapitaIncome_2005_Dollar_Analyze_Years.mat (Predefined input) T100_US_to_WorldRegion_Pax_Historical.mat

Output T100_US_to_NonUS_WorldRegion_Coef.mat % T100_US_to_NonUS_WorldRegion_Coef: Column 1 - Constant; % T100_US_to_NonUS_WorldRegion_Coef: Column 2 - Dummy_911; % T100_US_to_NonUS_WorldRegion_Coef: Column 3 - Dummy_911_2; % T100_US_to_NonUS_WorldRegion_Coef: Column 4 - SGDP; % T100_US_to_NonUS_WorldRegion_Coef: Column 5 - PGDP; % T100_US_to_NonUS_WorldRegion_Coef: Column 6 - S_GDP_per_Capita; % T100_US_to_NonUS_WorldRegion_Coef: Column 7 - P_GDP_per_Capita; % T100_US_to_NonUS_WorldRegion_Coef: Column 8 - R-Squared; % T100_US_to_NonUS_WorldRegion_Coef: Column 9 - Model Type (1-Linear/2-Log Linear);

234

Select_Preferred_Model.m % This function selects the preferred model from all the 8 candidate models % based on the defined weights for forecast growth rate and the R-Squared % value Input T100_US_to_NonUS_WorldRegion_Coef.mat

T100_US_to_WorldRegion_Pax_Historical.mat T100_US_to_WorldRegion_Pax_Forecast.mat W (Predefined variable) % Weight for R-Squared, Forecast_Growth_Rate_Near_Term, Forecast_Growth_Rate_Long_Term

Output Preferred_Model.mat Estimate_T100_US_to_WorldRegion_Pax_Analyze_Years.m % This function : % 1. Forecasts US to 9 NonUS WorldRegions by the econometric models % % 2. Forecasts US to 3 US WorldRegions (AK, HI and Others) by the average % growth factor of each US WorldRegions per year during historical yearsInput GDP_2005_Dollar_Analyze_Years.mat (Predefined input)

PerCapitaIncome_2005_Dollar_Analyze_Years.mat (Predefined input) T100_US_to_WorldRegion_Pax_Historical.mat T100_US_to_NonUS_WorldRegion_Coef.mat

Output Forecast_T100_US_to_NonUS_WorldRegion_Pax_Historical.mat T100_US_to_WorldRegion_Pax_Forecast.mat

Estimate_T100_US_Airport_to_WorldRegion_Pax_Forecast.m % This function : % 1. Summarizes the market share of each U.S. airport to each of 12 world % regions in base year % % 2. Estimate the passengers from each U.S. airport to each of 12 world % regions by distributing the total number of passengers from US to each % world region among the U.S. airports by the market share of each U.S. % airport in Step 1. Input T100_US_to_WorldRegion_Pax_Historical.mat

T100_US_to_WorldRegion_Pax_Forecast.mat T100_US_Airport_to_WorldRegion_Pax_Historical.mat

Output T100_US_Airport_to_WorldRegion_MarketShare_Base_Year.mat T100_US_Airport_to_WorldRegion_Pax_Forecast.mat

Filter_T100_US_Airport_to_WorldRegion_Pax.m % This function: % Filters the output 'US_Airport_to_WorldRegion_Pax' by the U.S. airport % from 'Focus_US_Airports' and combine all the other airports as 'Other'Input Focus_US_Airports.mat

T100_US_Airport_to_WorldRegion_Pax_Forecast.matOutput Focus_T100_US_Airport_to_WorldRegion_Pax_Forecast.mat

235

Plot_T100_US_to_WorldRegion_Pax_Historical_Model.m% This function plots the historical enplanements at U.S. to each world % region v.s. the corresponding set of forecast values by the model type % (Linear/LogLinear) of the forecast modelInput WorldRegion_Wac.mat (Predefined input)

T100_US_to_NonUS_WorldRegion_Coef.mat T100_US_to_WorldRegion_Pax_Historical.mat Forecast_T100_US_to_NonUS_WorldRegion_Pax_Historical.mat

Sample Output

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20081

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om U

.S.

5-Europe (2008 Market Share: 29.8228%) Yearly Passengers from U.S. 1990 - 2008

Forecast - Linear GDPUS+Europe (R2 = 0.97211)

Forecast - Linear GDPUS*Europe (R2 = 0.95464)


Forecast - Linear PerCapitaGDPUS*Europe (R2 = 0.95184)

Historical

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20081

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om U

.S.


Forecast - Log Linear GDPUS+Europe (R2 = 0.96348)

Forecast - Log Linear PGDPUS*Europe (R2 = 0.97)

Forecast - Log Linear PerCapitaGDPUS+Europe (R2 = 0.96825)

Forecast - Log Linear PerCapitaGDPUS*Europe (R2 = 0.95241)

Historical

236

Plot_T100_US_to_WorldRegion_Pax_Historical_Variable.m% This function plots the historical enplanements at U.S. to each world % region v.s. the corresponding set of forecast values by the indenpendent % variable (GDP/PerCapitaGDP)in the forecast modelInput WorldRegion_Wac.mat (Predefined input)

T100_US_to_NonUS_WorldRegion_Coef.mat T100_US_to_WorldRegion_Pax_Historical.mat Forecast_T100_US_to_NonUS_WorldRegion_Pax_Historical.mat

Sample Output

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20081

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om U

.S.


Forecast - Log Linear GDPUS+Europe (R2 = 0.96348)

Forecast - Log Linear PGDPUS*Europe (R2 = 0.97)

Forecast - Log Linear PerCapitaGDPUS+Europe (R2 = 0.96825)

Forecast - Log Linear PerCapitaGDPUS*Europe (R2 = 0.95241)

Historical

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20081

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om U

.S.


Forecast - Linear GDPUS+Europe (R2 = 0.97211)

Forecast - Linear GDPUS*Europe (R2 = 0.95464)


Forecast - Linear PerCapitaGDPUS*Europe (R2 = 0.95184)

Historical

237

Plot_T100_US_to_WorldRegion_Pax_Historical_Preferred.m% This function plots the historical enplanements at U.S. to each world % region v.s. the corresponding forecast values by the preferred modelInput WorldRegion_Wac.mat (Predefined input)

T100_US_to_NonUS_WorldRegion_Coef.mat T100_US_to_WorldRegion_Pax_Historical.mat Forecast_T100_US_to_NonUS_WorldRegion_Pax_Historical.mat Preferred_Model.mat

Sample Output

Plot_T100_US_to_WorldRegion_Pax_Forecast_Preferred.m% This function plots the historical and forecast enplanements at U.S. to % each world region during the whole analyze years by the preferred modelInput WorldRegion_Wac.mat (Predefined input)

T100_US_to_NonUS_WorldRegion_Coef.mat T100_US_to_WorldRegion_Pax_Historical.mat T100_US_to_WorldRegion_Pax_Forecast.mat Growth_Rate.mat Preferred_Model.mat

Sample Output

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 20081

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om U

.S.



Historical

1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20401

2

3

4

5

6

7

8x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om U

.S.




238

Plot_T100_US_to_WorldRegion_Pax_Forecast_Model.m% This function plots the historical and forecast enplanements at U.S. to % each world region during the whole analyze years by the model type % (Linear/LogLinear) of the forecast modelInput WorldRegion_Wac.mat (Predefined input)

T100_US_to_NonUS_WorldRegion_Coef.mat T100_US_to_WorldRegion_Pax_Historical.mat T100_US_to_WorldRegion_Pax_Forecast.mat Growth_Rate.mat

Sample Output

1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

2

4

6

8

10

12

14x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om U

.S.


Forecast - Linear GDPUS+Europe (R2 = 0.97211; Avg. Growth Factor = 3.76%)

Forecast - Linear GDPUS*Europe (R2 = 0.95464; Avg. Growth Factor = 5.2447%)


Forecast - Linear PerCapitaGDPUS*Europe (R2 = 0.95184; Avg. Growth Factor = 4.8049%)


1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

2

4

6

8

10

12

14

16

18x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om U

.S.


Forecast - Log Linear GDPUS+Europe (R2 = 0.96348; Avg. Growth Factor = 5.8583%)

Forecast - Log Linear PGDPUS*Europe (R2 = 0.97; Avg. Growth Factor = 5.7161%)

Forecast - Log Linear PerCapitaGDPUS+Europe (R2 = 0.96825; Avg. Growth Factor = 5.7992%)

Forecast - Log Linear PerCapitaGDPUS*Europe (R2 = 0.95241; Avg. Growth Factor = 6.0737%)


239

Plot_T100_US_to_WorldRegion_Pax_Forecast_Variable.m% This function plots the historical and forecast enplanements at U.S. to % each world region during the whole analyze years by the indenpendent % variable (GDP/PerCapitaGDP)in the forecast modelInput WorldRegion_Wac.mat (Predefined input)

T100_US_to_NonUS_WorldRegion_Coef.mat T100_US_to_WorldRegion_Pax_Historical.mat T100_US_to_WorldRegion_Pax_Forecast.mat Growth_Rate.mat

Sample Output

1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

2

4

6

8

10

12

14

16x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om U

.S.


Forecast - Linear GDPUS+Europe (R2 = 0.97211; Avg. Growth Factor = 3.76%)

Forecast - Linear GDPUS*Europe (R2 = 0.95464; Avg. Growth Factor = 5.2447%)

Forecast - Log Linear GDPUS+Europe (R2 = 0.96348; Avg. Growth Factor = 5.8583%)

Forecast - Log Linear PGDPUS*Europe (R2 = 0.97; Avg. Growth Factor = 5.7161%)


1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 20400

2

4

6

8

10

12

14

16

18x 10

7

Year

MA

RK

ET P

asse

nger

s fr

om U

.S.



Forecast - Linear PerCapitaGDPUS*Europe (R2 = 0.95184; Avg. Growth Factor = 4.8049%)

Forecast - Log Linear PerCapitaGDPUS+Europe (R2 = 0.96825; Avg. Growth Factor = 5.7992%)

Forecast - Log Linear PerCapitaGDPUS*Europe (R2 = 0.95241; Avg. Growth Factor = 6.0737%)


241

'Dest_State_Fips' '|' 'Dest_State' '|' 'Dest_State_Name' '|' 'Dest_Wac' '|' ... % Field 18-21 'Airport_Group' '|' 'Wac_Group' '|' 'Tk_Carrier_Change' '|' 'Tk_Carrier_Group' '|' ... % Field 22-25 'Op_Carrier_Change' '|' 'Op_Carrier_Group' '|' 'Rp_Carrier' '|' ... % Field 26-28 'Mkt_Tk_Carrier' '|' 'Mkt_Op_Carrier' '|' 'Bulk_Fare' '|' ... % Field 29-31 'Passengers' '|' 'Mkt_Fare' '|' 'Mkt_Distance' '|' 'Mkt_Distance_Group' '|' ... % Field 32-35 'Mkt_Miles_Flown' '|' 'NonStop_Miles' '|' 'Itin_Geo_Type' '|' 'Mkt_Geo_Type']; % Field 36-39 for Year = Historical_Years_Analyzed for Quarter = 1 : 4 FileName_Format_W = ['_Y', num2str(Year), '_Q', num2str(Quarter), '.txt']; if Year < 1998 % DB1A Fid_DB1A_D_Ticket = fopen([DB1A_Domestic_Dir, '\DB1A.D.Ticket', FileName_Format_W], 'w'); Fid_DB1A_D_Coupon = fopen([DB1A_Domestic_Dir, '\DB1A.D.Coupon', FileName_Format_W], 'w'); Fid_DB1A_D_Market = fopen([DB1A_Domestic_Dir, '\DB1A.D.Market', FileName_Format_W], 'w'); Fid_DB1A_I_Ticket = fopen([DB1A_International_Dir, '\DB1A.I.Ticket', FileName_Format_W], 'w'); Fid_DB1A_I_Coupon = fopen([DB1A_International_Dir, '\DB1A.I.Coupon', FileName_Format_W], 'w'); Fid_DB1A_I_Market = fopen([DB1A_International_Dir, '\DB1A.I.Market', FileName_Format_W], 'w'); fprintf(Fid_DB1A_D_Ticket,'%s\n',Title_Ticket); fprintf(Fid_DB1A_D_Coupon,'%s\n',Title_Coupon); fprintf(Fid_DB1A_D_Market,'%s\n',Title_Market); fprintf(Fid_DB1A_I_Ticket,'%s\n',Title_Ticket); fprintf(Fid_DB1A_I_Coupon,'%s\n',Title_Coupon); fprintf(Fid_DB1A_I_Market,'%s\n',Title_Market); % Recall the function 'DB1A_DB1B_Convert' to convert the DB1A raw data into Ticket, % Coupon and Market data, respectively, by itinerary geographic type % 0/1 - International/non-continental % 2 - Domestic(continental) DB1A_DB1B_Convert(Year,Quarter, DB1A_Raw_Dir, DB1B_Raw_Dir, ... Fid_DB1A_D_Ticket, Fid_DB1A_D_Coupon, Fid_DB1A_D_Market, ... Fid_DB1A_I_Ticket, Fid_DB1A_I_Coupon, Fid_DB1A_I_Market); else % DB1B Fid_DB1B_D_Ticket = fopen([DB1B_Domestic_Dir, '\DB1B.D.Ticket', FileName_Format_W], 'w'); Fid_DB1B_D_Coupon = fopen([DB1B_Domestic_Dir, '\DB1B.D.Coupon', FileName_Format_W], 'w'); Fid_DB1B_D_Market = fopen([DB1B_Domestic_Dir, '\DB1B.D.Market', FileName_Format_W], 'w'); Fid_DB1B_I_Ticket = fopen([DB1B_International_Dir, '\DB1B.I.Ticket', FileName_Format_W], 'w'); Fid_DB1B_I_Coupon = fopen([DB1B_International_Dir, '\DB1B.I.Coupon', FileName_Format_W], 'w');

242

Fid_DB1B_I_Market = fopen([DB1B_International_Dir, '\DB1B.I.Market', FileName_Format_W], 'w'); fprintf(Fid_DB1B_D_Ticket,'%s\n',Title_Ticket); fprintf(Fid_DB1B_D_Coupon,'%s\n',Title_Coupon); fprintf(Fid_DB1B_D_Market,'%s\n',Title_Market); fprintf(Fid_DB1B_I_Ticket,'%s\n',Title_Ticket); fprintf(Fid_DB1B_I_Coupon,'%s\n',Title_Coupon); fprintf(Fid_DB1B_I_Market,'%s\n',Title_Market); % Recall the function 'DB1A_DB1B_Convert' to convert the DB1A/DB1B raw data into Ticket, % Coupon and Market data, respectively, by itinerary geographic type % 0/1 - International/non-continental % 2 - Domestic(continental) DB1A_DB1B_Convert(Year,Quarter, DB1A_Raw_Dir, DB1B_Raw_Dir, ... Fid_DB1B_D_Ticket, Fid_DB1B_D_Coupon, Fid_DB1B_D_Market, ... Fid_DB1B_I_Ticket, Fid_DB1B_I_Coupon, Fid_DB1B_I_Market); end % if Year < 1998 % DB1A end % for Quarter = 1 : 4 end % for Year = Historical_Years_Analyzed return

function DB1A_DB1B_Convert(Year,Quarter,Input_Dir_DB1A,Input_Dir_DB1B, ... Fid_D_Ticket,Fid_D_Coupon,Fid_D_Market, ... Fid_I_Ticket,Fid_I_Coupon,Fid_I_Market) %-------------------------------------------------------------------------- % This function converts the DB1A/DB1B raw data into Ticket, Coupon and % Market data, respectively, by itinerary geographic type % 0/1 - International/non-continental % 2 - Domestic(continental) % (The raw data dictionary is based on the document titled % 'BTS Database DB1A Record Generation (04/16/2003)' % % Calling: % N/A % % Called: % Main_DB1A_DB1B_Convert % % Revised by N. Shen (May 26, 2009) %-------------------------------------------------------------------------- % DB1B raw data dictionary % Field 1: Dollar_Value % Field 2: Rp_Carrier % Field 3: Date % Field 4: Coupon_Segments % Field 5: Num_Of_Pax % Field 6: Origin_Airport_Code % Field 7: Dollar_Credibility % Field 8: City_Numeric_Code % Field 9: Airport_Ind % Field 10: City_World_Area

243

% Field 11: Op_Carrier_Code; Repeat 1 % Field 12: Coupon_Type_Code_Op; Repeat 2 % Field 13: Ticketed_Carrier_Code; Repeat 3 % Field 14: Coupon_Type_Code_Tk; Repeat 4 % Field 15: Fare_Class_Code; Repeat 5 % Field 16: Coupon_Distance; Repeat 6 % Field 17: Airport_Code; Repeat 7 % Field 18: Trip_Break_Code; Repeat 8 % Field 19: City_Numeric_Code2; Repeat 9 % Field 20: Airport_Ind2 Repeat 10 % Field 21: City_World_Area2; Repeat 11 Col_Dollar_Value = 1; Col_Rp_Carrier = 2; Col_Date = 3; Col_Coupon_Segments = 4; Col_Num_Of_Passengers = 5; Col_Origin_Airport_Code = 6; Col_Dollar_Credibility_Ind = 7; Col_City_Numeric_Code = 8; Col_Airport_Ind = 9; Col_City_World_Area = 10; Col_Op_Carrier_Code = 11; if Year < 1998 % DB1A raw data input FileName_Format_R = ['_Y', num2str(Year), '_Q', num2str(Quarter), '.asc']; Fid_Input = fopen([Input_Dir_DB1A, '\DB1A', FileName_Format_R], 'r'); Col_Repeat = 9; Format_Repeat = '%s%s%s%f%s%s%f%f%f'; Col_Coupon_Type_Code_Operating = 13; Col_Tk_Carrier = []; Col_Fare_Class_Code = 12; Col_Coupon_Distance = 14; Col_Airport_Code = 15; Col_Trip_Break_Code = 16; Col_City_World_Area2 = 19; else % % DB1B raw data input if Quarter == 4 FileName_Format_R = [num2str(Year), num2str(Quarter*3), '.asc']; else FileName_Format_R = [num2str(Year), '0', num2str(Quarter*3), '.asc']; end % for Quarter == 4 Fid_Input = fopen([Input_Dir_DB1B, '\DB1B.', FileName_Format_R], 'r'); Col_Repeat = 11; Format_Repeat = '%s%s%s%s%s%f%s%s%f%f%f'; Col_Coupon_Type_Code_Operating = 12; Col_Tk_Carrier = 13; Col_Fare_Class_Code = 15; Col_Coupon_Distance = 16; Col_Airport_Code = 17; Col_Trip_Break_Code = 18; Col_City_World_Area2 = 21; end % if Year < 1998 % DB1A data Record_Sequence = 0; % Cumulative count of raw DB1B data records

244

Record_Sequence_D = 0; % Cumulative count of domestic DB1B data records Record_Sequence_I = 0; % Cumulative count of international DB1B data records Mkt_Sequence = 1; % Cumulative count of markets in each DB1B data record tic; while feof(Fid_Input) == 0 Format = '%f%s%f%f%f%s%s%f%f%f'; Record_Sequence = Record_Sequence + 1; if rem(Record_Sequence,10000) == 0 disp(['Elapsed time is ' num2str(toc/60) ' minutes']) disp(['Records Completed = ', num2str(Record_Sequence)]) disp(['Year = ', num2str(Year)]) end % if rem(Record_Sequence,10000) == 0 Line = fgets(Fid_Input); Raw_Data_Short = textscan(Line, '%f%s%f%f%*[^\n]', 'Delimiter', '|'); Coupons = Raw_Data_Short{Col_Coupon_Segments}; Step = (Coupons-1)*Col_Repeat; for i = 1 : Coupons Format = [Format Format_Repeat]; end % for i = 1 : Coupons Raw_Data = textscan(Line, Format, 'Delimiter', '|', 'EmptyValue', -Inf); % Itinerary information from DB1A/DB1B_Raw data Dollar_Value = Raw_Data{Col_Dollar_Value}; Rp_Carrier = Raw_Data{Col_Rp_Carrier}; Date = Raw_Data{Col_Date}; Passengers = Raw_Data{Col_Num_Of_Passengers}; Dollar_Credibility = Raw_Data{Col_Dollar_Credibility_Ind}; % Coupon information from DB1A/DB1B_Raw data Origin = Raw_Data(Col_Origin_Airport_Code:Col_Repeat:(Col_Origin_Airport_Code+Step)); Origin_Wac = Raw_Data(Col_City_World_Area:Col_Repeat:(Col_City_World_Area+Step)); Origin_City_Num = Raw_Data(Col_City_Numeric_Code:Col_Repeat:(Col_City_Numeric_Code+Step)); Origin_Apt_Ind = Raw_Data(Col_Airport_Ind:Col_Repeat:(Col_Airport_Ind+Step)); Op_Carrier = Raw_Data(Col_Op_Carrier_Code:Col_Repeat:(Col_Op_Carrier_Code+Step)); Coupon_Type = Raw_Data(Col_Coupon_Type_Code_Operating:Col_Repeat:(Col_Coupon_Type_Code_Operating+Step)); if Year < 1998 % DB1A data Tk_Carrier = cell(1,Coupons); % DB1A data doesn't include this field. Empty cell is assigned. else % DB1B data Tk_Carrier = Raw_Data(Col_Tk_Carrier:Col_Repeat:(Col_Tk_Carrier+Step)); end % if Year < 1998 % DB1A data Fare_Class = Raw_Data(Col_Fare_Class_Code:Col_Repeat:(Col_Fare_Class_Code+Step)); Trip_Break = Raw_Data(Col_Trip_Break_Code:Col_Repeat:(Col_Trip_Break_Code+Step)); Distance = Raw_Data(Col_Coupon_Distance:Col_Repeat:(Col_Coupon_Distance+Step)); Dest = Raw_Data(Col_Airport_Code:Col_Repeat:(Col_Airport_Code+Step)); Dest_Wac = Raw_Data(Col_City_World_Area2:Col_Repeat:(Col_City_World_Area2+Step));

245

Dest_City_Num = Raw_Data(Col_City_Numeric_Code:Col_Repeat:(Col_City_Numeric_Code+Step)); Dest_Apt_Ind = Raw_Data(Col_Airport_Ind:Col_Repeat:(Col_Airport_Ind+Step)); Origin_State = cell(1,Coupons); Origin_State_Name = cell(1,Coupons); Origin_State_Fips = cell(1,Coupons); Origin_Country = cell(1,Coupons); Dest_State = cell(1,Coupons); Dest_State_Name = cell(1,Coupons); Dest_State_Fips = cell(1,Coupons); Dest_Country = cell(1,Coupons); % -- 1st Converstion --: DB1A/DB1B_Ticket % Field 01|02|03: Itin_ID|Year|Quarter % Field 04|05|06|07: Rp_Carrier|Origin|Origin_City_Num|Origin_Wac % Field 08|09|10|11: Origin_Apt_Ind|Origin_State|Origin_State_Fips|Origin_Country % Field 12|13: Passengers|Coupons % Field 14: Itin_Geo_Type % International: Itin_Geo_Type = 0 % Non-continental: Itin_Geo_Type = 1 % Continental: Itin_Geo_Type = 2 for i = 1 : Coupons Itin_Geo_Type = (Origin_Wac{i} <= 99 & Dest_Wac{i} <= 99); if Itin_Geo_Type == 0 % International break else % Continental or Non-continental Itin_Geo_Type = 1 + (Origin_Wac{i} > 5 & Dest_Wac{i} > 5); % Non-continental if Itin_Geo_Type == 1 % Non-continental break end % if Itin_Geo_Type == 1 % Non-continental end % if Domestic_Trip_ID == 0 % International or Non-continental end % for i = 1 : Coupons if Itin_Geo_Type == 2 % Continental Record_Sequence_D = Record_Sequence_D + 1; Itin_ID = [num2str(Date) num2str(Record_Sequence_D)]; Fid_Ticket = Fid_D_Ticket; Fid_Coupon = Fid_D_Coupon; Fid_Market = Fid_D_Market; else % International or Non-Continental Record_Sequence_I = Record_Sequence_I + 1; Itin_ID = [num2str(Date) num2str(Record_Sequence_I)]; Fid_Ticket = Fid_I_Ticket; Fid_Coupon = Fid_I_Coupon; Fid_Market = Fid_I_Market; end % if Itin_Geo_Type == 2 % Continental % Field 15: RoundTrip RoundTrip = strcmp(Origin{1},Dest{end}); % Field 16: Online Online = 1; for i = 1 : (Coupons-1)

246

Online = strcmp(Op_Carrier{i},Op_Carrier{i+1}); if Online == 0 break end % if Online == 0 end % for i = 1 : (Coupons-1) % Field 17: Dollar_Cred Dollar_Cred = strcmp(Dollar_Credibility, ''); % Field 18: Itin_Fare if Dollar_Value == 99999 Itin_Fare = 0; else Itin_Fare = Dollar_Value; end % if Dollar_Value == 99999 % Field 19|20|21|22: Itin_Yield|Itin_Distance|Itin_Distance_Group|Itin_Miles_Flown Itin_Distance = 0; Itin_Miles_Flown = 0; for i = 1 : Coupons Itin_Distance = Itin_Distance + Distance{i}; if strcmp(Op_Carrier{i},'--') == 0 Itin_Miles_Flown = Itin_Miles_Flown + Distance{i}; end % if strcmp(Op_Carrier{i},'--') == 0 end % for i = 1 : Coupons if Dollar_Value == 99999 || Itin_Miles_Flown == 0 Itin_Yield = 0; else Itin_Yield = Dollar_Value / Itin_Miles_Flown; end % if Dollar_Value == 99999 || Miles_Flown == 0; Itin_Distance_Group = ceil(Itin_Distance/500); % Field 23: Bulk_Fare if Dollar_Value == 99999 Bulk_Fare = 1; else Bulk_Fare = 0; end % if Dollar_Value == 99999 % Field 24: Itin_Gateway for i = 1 : Coupons Condition1 = ((Origin_Wac{i} > 99) & (Dest_Wac{i} <= 99)); Condition2 = ((Origin_Wac{i} <= 99) & (Dest_Wac{i} > 99)); Itin_Gateway = (Condition1 | Condition2); if Itin_Gateway == 1 break end % if Itin_Gateway == 1 end % for i = 1 : Coupons NewLine_Ticket = ... [Itin_ID '|' num2str(Coupons) '|' num2str(Year) '|' num2str(Quarter) '|' ... % Field 01-04

247

char(Origin{1}) '|' num2str(Origin_Apt_Ind{1}) '|' num2str(Origin_City_Num{1}) '|' char(Origin_Country{1}) '|' ... % Field 05-08 num2str(Origin_State_Fips{1}) '|' char(Origin_State{1}) '|' char(Origin_State_Name{1}) '|' num2str(Origin_Wac{1}) '|' ... % Field 09-12 num2str(RoundTrip) '|' num2str(Online) '|' num2str(Dollar_Cred) '|' num2str(Itin_Yield) '|' ... % Field 13-16 char(Rp_Carrier) '|' num2str(Passengers) '|' num2str(Itin_Fare) '|' num2str(Bulk_Fare) '|' ... % Field 17-20 num2str(Itin_Distance) '|' num2str(Itin_Distance_Group) '|' num2str(Itin_Miles_Flown) '|' num2str(Itin_Geo_Type)]; % Field 21-24 fprintf(Fid_Ticket,'%s\n',NewLine_Ticket); % -- 2nd Converstion --: DB1A/DB1B_Coupon % Field 01|02|03|04|05: Itin_ID|Seq_Num|Mkt_ID|Year|Quarter Mkt_ID = [num2str(Date) num2str(Mkt_Sequence)]; % Field 06|07|08: Rp_Carrier|Tk_Carrier|Op_Carrier % Field 09|10|11|12: Origin|Origin_City_Num|Origin_Wac|Origin_Apt_Ind % Field 13|14|15: Origin_State|Origin_State_Fips|Origin_Country % Field 16|17|18|19: Dest|Dest_City_Num|Dest_Wac|Dest_Apt_Ind % Field 20|21|22: Dest_State|Dest_State_Fips|Dest_Country % Field 23|24|25|26: Passengers|Coupons|Coupon_Type|Itin_Geo_Type % Field 27: Coupon_Geo_Type % International: Coupon_Geo_Type = 0 % Non-continental: Coupon_Geo_Type = 1 % Continental: Coupon_Geo_Type = 2 Coupon_Geo_Type = zeros(1,Coupons); for i = 1 : Coupons Coupon_Geo_Type(i) = (Origin_Wac{i} <= 99 & Dest_Wac{i} <= 99); if Coupon_Geo_Type(i) ~= 0 % Continental or Non-continental Coupon_Geo_Type(i) = 1 + (Origin_Wac{i} > 5 & Dest_Wac{i} > 5); end % if Coupon_Geo_Type(i) ~= 2 % International or Non-continental end % for i = 1 : Coupons % Field 28|29: Fare_Class|Trip_Break % Field 30/31/32: Gateway/Distance/Distance_Group Gateway = zeros(1,Coupons); Distance_Group = zeros(1,Coupons); for i = 1 : Coupons Condition1 = ((Origin_Wac{i} > 99) & (Dest_Wac{i} <= 99)); Condition2 = ((Origin_Wac{i} <= 99) & (Dest_Wac{i} > 99)); Gateway(i) = (Condition1 | Condition2); Distance_Group(i) = ceil(Distance{i}/500); end % for i = 1 : Coupons j = 1; % j = Origin coupon for i = 1 : Coupons NewLine_Coupon = ... [Itin_ID '|' Mkt_ID '|' num2str(i) '|' num2str(Coupons) '|' num2str(Year) '|' num2str(Quarter) '|' ... % Field 01-06 char(Origin{i}) '|' num2str(Origin_Apt_Ind{i}) '|' num2str(Origin_City_Num{i}) '|' char(Origin_Country{i}) '|' ... % Field 07-10

248

num2str(Origin_State_Fips{i}) '|' char(Origin_State{i}) '|' char(Origin_State_Name{i}) '|' num2str(Origin_Wac{i}) '|' ... % Field 11-14 char(Dest{i}) '|' num2str(Dest_Apt_Ind{i}) '|' num2str(Dest_City_Num{i}) '|' char(Dest_Country{i}) '|' ... % Field 15-18 num2str(Dest_State_Fips{i}) '|' char(Dest_State{i}) '|' char(Dest_State_Name{i}) '|' num2str(Dest_Wac{i}) '|' ... % Field 19-22 char(Trip_Break{i}) '|' char(Coupon_Type{i}) '|' char(Tk_Carrier{i}) '|' char(Op_Carrier{i}) '|' ... % Field 23-26 char(Rp_Carrier) '|' num2str(Passengers) '|' char(Fare_Class{i}) '|' num2str(Distance{i}) '|' ... % Field 27-30 num2str(Distance_Group(i)) '|' num2str(Gateway(i)) '|' num2str(Itin_Geo_Type) '|' num2str(Coupon_Geo_Type(i))]; % Field 31-34 fprintf(Fid_Coupon,'%s\n',NewLine_Coupon); if strcmp(Trip_Break{i},'X') == 1 % -- 3rd Converstion --: DB1A/DB1B_Market % Field 01|02|03|04|05: Itin_ID|Mkt_ID|Year|Quarter|Rp_Carrier % Field 06|07|08|09: Origin|Origin_City_Num|Origin_Wac|Origin_Apt_Ind % Field 10|11|12: Origin_State|Origin_State_Fips|Origin_Country % Field 13|14|15|16: Dest|Dest_City_Num|Dest_Wac|Dest_Apt_Ind % Field 17|18|19: Dest_State|Dest_State_Fips|Dest_Country % Field 20|21|22: Passengers|Mkt_Coupons|Itin_Geo_Type Mkt_Coupons = size(j:i,2); % Field 23: Mkt_Geo_Type for ii = j : i Mkt_Geo_Type = ((Origin_Wac{ii} <= 99) & (Dest_Wac{ii} <= 99)); if Mkt_Geo_Type == 0 % International break else % Continental or Non-continental Mkt_Geo_Type = 1 + (Origin_Wac{ii} > 5 & Dest_Wac{ii} > 5); % 1: Non-continental; 2: Continental if Mkt_Geo_Type == 1 % Non-continental break end % if Mkt_Geo_Type == 1 end % if Mkt_Geo_Type == 0 end % for ii = j : i % Field 24|25|26: Mkt_Fare|Mkt_Distance|Mkt_Miles_Flown Mkt_Distance = 0; Mkt_Miles_Flown = 0; for k = j:i Mkt_Distance = Mkt_Distance + Distance{k}; if strcmp(Op_Carrier{k},'--') == 0 Mkt_Miles_Flown = Mkt_Miles_Flown + Distance{k}; end % if strcmp(Op_Carrier_Code{i},'--') == 0 end % for i = 1 : Mkt_Coupons if Bulk_Fare == 1 Mkt_Fare = 0; else Mkt_Fare = Itin_Yield * Mkt_Miles_Flown; end % if Bulk_Fare == 1

249

% Field 27: NonStop_Miles NonStop_Miles = 0; % Field 28: Mkt_Distance_Group Mkt_Distance_Group = ceil(Mkt_Distance/500); % Field 29|30|31|32: Op_Carrier_Group|Tk_Carrier_Group|Airport_Group|Wac_Group Op_Carrier_Group = char(Op_Carrier{j}); Tk_Carrier_Group = char(Tk_Carrier{j}); Airport_Group = char(Origin{j}); Wac_Group = num2str(Origin_Wac{j}); if j+1 <= i for k = (j+1):i Op_Carrier_Group = [Op_Carrier_Group ':' char(Op_Carrier{k})]; Tk_Carrier_Group = [Tk_Carrier_Group ':' char(Tk_Carrier{k})]; Airport_Group = [Airport_Group ':' char(Origin{k})]; Wac_Group = [Wac_Group ':' num2str(Origin_Wac{k})]; end % for k = j:i end % if j+1 <= i Airport_Group = [Airport_Group ':' char(Dest{i})]; Wac_Group = [Wac_Group ':' num2str(Dest_Wac{i})]; % Field 33|34: Op_Carrier_Change|Tk_Carrier_Change if j+1 <= i for k = (j+1):i Op_Carrier_Change = 1 - strcmp(Op_Carrier{k},Op_Carrier{j}); if Op_Carrier_Change == 1 break end % if Op_Carrier_Change == 1 end % for k = j:i for k = (j+1):i Tk_Carrier_Change = 1 - strcmp(Tk_Carrier{k},Tk_Carrier{j}); if Tk_Carrier_Change == 1 break end % if Tk_Carrier_Change == 1 end % for k = j:i else Op_Carrier_Change = 0; Tk_Carrier_Change = 0; end % if j+1 <= i % Field 35|36: Origin_State_NM|Dest_State_NM % Field 37: Mkt_Tk_Carrier if Tk_Carrier_Change == 0 Mkt_Tk_Carrier = Tk_Carrier{j}; else Mkt_Tk_Carrier = '99'; end % if Op_Carrier_Change == 0 % Field 38: Mkt_Op_Carrier if Op_Carrier_Change == 0 Mkt_Op_Carrier = Op_Carrier{j}; else

250

Mkt_Op_Carrier = '99'; end % if Op_Carrier_Change == 0 % Field 39: Bulk_Fare; NewLine_Market = ... [Itin_ID '|' Mkt_ID '|' num2str(Mkt_Coupons) '|' num2str(Year) '|' num2str(Quarter) '|' ... % Field 01-05 char(Origin{j}) '|' num2str(Origin_Apt_Ind{j}) '|' num2str(Origin_City_Num{j}) '|' char(Origin_Country{j}) '|' ... % Field 06-09 num2str(Origin_State_Fips{j}) '|' char(Origin_State{j}) '|' char(Origin_State_Name{j}) '|' num2str(Origin_Wac{j}) '|' ... % Field 10-13 char(Dest{i}) '|' num2str(Dest_Apt_Ind{i}) '|' num2str(Dest_City_Num{i}) '|' char(Dest_Country{i}) '|' ... % Field 14-17 num2str(Dest_State_Fips{i}) '|' char(Dest_State{i}) '|' char(Dest_State_Name{i}) '|' num2str(Dest_Wac{i}) '|' ... % Field 18-21 Airport_Group '|' Wac_Group '|' num2str(Tk_Carrier_Change) '|' Tk_Carrier_Group '|' ... % Field 22-25 num2str(Op_Carrier_Change) '|' Op_Carrier_Group '|' char(Rp_Carrier) '|' ... % Field 26-28 char(Mkt_Tk_Carrier) '|' char(Mkt_Op_Carrier) '|' num2str(Bulk_Fare) '|' ... % Field 29-31 num2str(Passengers) '|' num2str(Mkt_Fare) '|' num2str(Mkt_Distance) '|' num2str(Mkt_Distance_Group) '|' ... % Field 32-35 num2str(Mkt_Miles_Flown) '|' num2str(NonStop_Miles) '|' num2str(Itin_Geo_Type) '|' num2str(Mkt_Geo_Type)]; % Field 36-39 fprintf(Fid_Market,'%s\n',NewLine_Market); Mkt_Sequence = Mkt_Sequence + 1; Mkt_ID = [num2str(Date) num2str(Mkt_Sequence)]; j = i + 1; end % if strcmp(Trip_Break{i},'X') == 1 end % for i = 1 : Coupons end % while feof(Fid_Input) == 0 fclose('all') return

251

Appendix C: The Impact of the EU-US Open Skies Agreement on Commercial Airline

Passenger Traffic over the North Atlantic

252

C.1 Passenger Traffic Symmetry Observed at the Top 91 Airport Pairs between the United States and the

Selected Nine European Countries

Figure C-1: Airport-to-Airport Passenger Traffic from Europe to the United States vs. Traffic from the United

States to Europe (Source: 2007 T100 International Market Data).

0.0 0.5 1.0

JFK:LHR

LAX:LHR

IAD:LHR

BOS:LHR

ORD:FRA

DTW:AMS

JFK:FRA

SFO:FRA

IAD:CDG

LAX:CDG

EWR:CDG

JFK:DUB

JFK:MAD

EWR:FRA

DTW:FRA

ATL:LGW

LAX:FRA

ORD:CDG

JFK:ZRH

PHL:FRA

MCO:MAN

DFW:FRA

EWR:LGW

ORD:MUC

ORD:AMS

PHL:CDG

DTW:CDG

BOS:FRA

LAX:AMS

JFK:SNN

MIA:FRA

DEN:FRA

SEA:LHR

EWR:MAN

SFO:MUC

ATL:FCO

EWR:MAD

JFK:MUC

DTW:LGW

BOS:DUB

LAX:MUC

PHX:LHR

SEA:AMS

PHL:MAN

CLT:LGW

Passengers for Top 91 Transatlantic Airport Pairs (Millions)

Europe to U.S.U.S. to Europe

The symmetry is present for airportpairs with the top 91 passenger traffic in 2007 . These airport pairs captured 80% of total passenger traffic from 31 U.S. airports to 35 European airports in analysis domain.

253

C.2 Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the United States to Each of

the Selected Nine European Countries

Figure C-2: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to the U.K.


Figure C-3: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to Germany


1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 20200.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2 x 107 1-U.K. (2007 Market Share: 33.4%) Yearly Passengers from U.S. 1990 - 2020

Year

Pass

enge

rs fr

om U

.S.


1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 20202

3

4

5

6

7

8

9 x 106 2-Germany (2007 Market Share: 18.4%) Yearly Passengers from U.S. 1990 - 2020

Year

Pass

enge

rs fr

om U

.S.


254

Figure C-4: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to France


Figure C-5: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to Netherlands


1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 20201

2

3

4

5

6

7 x 106 3-France (2007 Market Share: 11.8%) Yearly Passengers from U.S. 1990 - 2020

Year

Pass

enge

rs fr

om U

.S.


1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 20200

1

2

3

4

5

6

7

8

9 x 106 4-Netherlands (2007 Market Share: 9%) Yearly Passengers from U.S. 1990 - 2020

Year

Pass

enge

rs fr

om U

.S.


255

Figure C-6: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to Italy


Figure C-7: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to Ireland


1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 20200.5

1

1.5

2

2.5

3 x 106 5-Italy (2007 Market Share: 5.7%) Yearly Passengers from U.S. 1990 - 2020

Year

Pass

enge

rs fr

om U

.S.


1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 20200

0.5

1

1.5

2

2.5 x 106 6-Ireland (2007 Market Share: 4.2%) Yearly Passengers from U.S. 1990 - 2020

Year

Pass

enge

rs fr

om U

.S.


256

Figure C-8: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to Spain


Figure C-9: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to Switzerland


1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 20200.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2 x 106 7-Spain (2007 Market Share: 3.9%) Yearly Passengers from U.S. 1990 - 2020

Year

Pass

enge

rs fr

om U

.S.


1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 20200.4

0.6

0.8

1

1.2

1.4

1.6

1.8 x 106 8-Switzerland (2007 Market Share: 3%) Yearly Passengers from U.S. 1990 - 2020

Year

Pass

enge

rs fr

om U

.S.


257

Figure C-10: Historical (1990 – 2007) and Forecast (2008 – 2020) Enplanements from the U.S. to Belgium


1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 20202

3

4

5

6

7

8

9

10

11

12 x 105 9-Belgium (2007 Market Share: 1.6%) Yearly Passengers from U.S. 1990 - 2020

Year

Pass

enge

rs fr

om U

.S.


258

Table C-1: Passenger Traffic between the United States and the Selected Nine European Countries in Analysis Domain during 1990 – 2020

(1990 – 2007 Data Source: 1990 – 2007 T100 International Market Data).

1 2 3 4 5 6 7 8 9U.K. Germany France Netherlands Italy Ireland Spain Switzerland Belgium

1990 4,528,446 2,256,382 1,470,630 637,325 669,932 279,364 468,842 574,408 342,196 11,227,5251991 4,098,713 2,156,690 1,397,422 687,294 626,814 231,550 439,215 503,685 275,476 10,416,8591992 4,842,188 2,555,764 1,671,992 776,632 840,132 258,562 578,701 542,098 297,750 12,363,8191993 5,174,786 2,657,348 1,606,463 964,422 822,010 248,994 519,705 581,136 335,986 12,910,8501994 5,333,169 2,690,340 1,738,517 1,094,743 882,644 316,378 577,783 628,521 345,132 13,607,2271995 5,759,648 2,845,646 1,765,418 1,186,835 920,484 338,800 605,000 680,574 366,790 14,469,1951996 6,141,315 2,912,320 1,949,248 1,404,848 953,423 416,048 567,609 729,683 389,560 15,464,0541997 6,829,973 3,151,366 2,056,891 1,652,792 1,023,953 414,485 619,862 774,321 492,387 17,016,0301998 7,606,402 3,247,833 2,208,808 2,000,085 1,009,600 484,939 690,228 885,156 613,387 18,746,4381999 8,206,666 3,504,471 2,369,384 2,189,654 1,148,349 630,692 767,081 990,856 695,452 20,502,6052000 8,665,584 3,723,927 3,008,111 2,265,435 1,461,210 748,719 800,155 1,050,556 728,523 22,452,2202001 7,660,202 3,338,407 2,878,633 2,017,044 1,202,177 743,481 682,384 909,585 575,957 20,007,8702002 7,692,186 3,277,681 2,706,402 1,974,929 1,001,086 639,532 653,740 717,803 260,257 18,923,6162003 7,552,207 3,403,578 2,556,152 1,959,699 950,437 774,089 653,147 698,051 261,028 18,808,3882004 8,216,281 3,807,120 2,779,885 2,139,464 1,205,983 829,352 765,522 704,368 331,242 20,779,2172005 8,344,689 4,093,115 2,986,896 2,272,816 1,277,667 909,869 811,046 699,668 363,961 21,759,7272006 8,276,607 4,285,164 2,973,593 2,238,902 1,314,422 996,896 858,197 720,786 360,237 22,024,8042007 8,513,976 4,677,240 3,013,163 2,344,010 1,443,529 1,074,759 1,003,373 794,974 396,021 23,261,0452008 9,029,528 4,899,345 3,183,732 2,600,525 1,521,776 1,152,753 1,046,962 844,274 426,803 24,705,6982009 9,605,373 5,125,107 3,370,979 2,900,243 1,606,402 1,247,148 1,094,419 898,027 463,158 26,310,8562010 10,237,314 5,356,814 3,576,641 3,232,907 1,700,217 1,342,442 1,146,109 955,523 504,585 28,052,5522011 10,907,512 5,599,758 3,793,867 3,599,033 1,799,697 1,437,451 1,200,505 1,013,503 548,699 29,900,0252012 11,618,282 5,854,483 4,023,303 4,001,955 1,905,181 1,536,199 1,257,753 1,076,470 595,919 31,869,5452013 12,372,078 6,121,560 4,265,637 4,445,376 2,017,031 1,636,877 1,318,002 1,142,410 648,592 33,967,5632014 13,171,511 6,401,586 4,521,597 4,933,397 2,135,635 1,740,706 1,381,408 1,213,729 707,316 36,206,8852015 14,004,256 6,688,770 4,786,483 5,459,060 2,258,508 1,847,467 1,446,955 1,288,161 766,714 38,546,3742016 14,886,509 6,989,574 5,065,958 6,036,428 2,388,625 1,956,868 1,515,875 1,362,265 830,453 41,032,5552017 15,821,223 7,304,644 5,360,830 6,670,646 2,526,418 2,074,924 1,588,343 1,441,522 898,819 43,687,3692018 16,811,516 7,634,659 5,671,944 7,367,252 2,672,337 2,193,472 1,664,542 1,524,497 972,721 46,512,9402019 17,860,696 7,980,328 6,000,192 8,132,382 2,826,863 2,318,859 1,744,665 1,613,046 1,053,598 49,530,6292020 18,972,251 8,342,391 6,346,523 8,972,848 2,990,502 2,451,478 1,828,911 1,705,779 1,139,359 52,750,042

Year\European CountryPassengers

Historical

Forecast

Sum

259

Table C-2: Growth of Passenger Traffic between the United States and the Selected Nine European Countries in Analysis Domain during 1990

– 2020 (1990 – 2007 Data Source: 1990 – 2007 T100 International Market Data).


1990 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐1991 ‐9.5% ‐4.4% ‐5.0% 7.8% ‐6.4% ‐17.1% ‐6.3% ‐12.3% ‐19.5% ‐7.2%1992 18.1% 18.5% 19.6% 13.0% 34.0% 11.7% 31.8% 7.6% 8.1% 18.7%1993 6.9% 4.0% ‐3.9% 24.2% ‐2.2% ‐3.7% ‐10.2% 7.2% 12.8% 4.4%1994 3.1% 1.2% 8.2% 13.5% 7.4% 27.1% 11.2% 8.2% 2.7% 5.4%1995 8.0% 5.8% 1.5% 8.4% 4.3% 7.1% 4.7% 8.3% 6.3% 6.3%1996 6.6% 2.3% 10.4% 18.4% 3.6% 22.8% ‐6.2% 7.2% 6.2% 6.9%1997 11.2% 8.2% 5.5% 17.6% 7.4% ‐0.4% 9.2% 6.1% 26.4% 10.0%1998 11.4% 3.1% 7.4% 21.0% ‐1.4% 17.0% 11.4% 14.3% 24.6% 10.2%1999 7.9% 7.9% 7.3% 9.5% 13.7% 30.1% 11.1% 11.9% 13.4% 9.4%2000 5.6% 6.3% 27.0% 3.5% 27.2% 18.7% 4.3% 6.0% 4.8% 9.5%2001 ‐11.6% ‐10.4% ‐4.3% ‐11.0% ‐17.7% ‐0.7% ‐14.7% ‐13.4% ‐20.9% ‐10.9%2002 0.4% ‐1.8% ‐6.0% ‐2.1% ‐16.7% ‐14.0% ‐4.2% ‐21.1% ‐54.8% ‐5.4%2003 ‐1.8% 3.8% ‐5.6% ‐0.8% ‐5.1% 21.0% ‐0.1% ‐2.8% 0.3% ‐0.6%2004 8.8% 11.9% 8.8% 9.2% 26.9% 7.1% 17.2% 0.9% 26.9% 10.5%2005 1.6% 7.5% 7.4% 6.2% 5.9% 9.7% 5.9% ‐0.7% 9.9% 4.7%2006 ‐0.8% 4.7% ‐0.4% ‐1.5% 2.9% 9.6% 5.8% 3.0% ‐1.0% 1.2%2007 2.9% 9.1% 1.3% 4.7% 9.8% 7.8% 16.9% 10.3% 9.9% 5.6%2008 6.1% 4.7% 5.7% 10.9% 5.4% 7.3% 4.3% 6.2% 7.8% 6.2%2009 6.4% 4.6% 5.9% 11.5% 5.6% 8.2% 4.5% 6.4% 8.5% 6.5%2010 6.6% 4.5% 6.1% 11.5% 5.8% 7.6% 4.7% 6.4% 8.9% 6.6%2011 6.5% 4.5% 6.1% 11.3% 5.9% 7.1% 4.7% 6.1% 8.7% 6.6%2012 6.5% 4.5% 6.0% 11.2% 5.9% 6.9% 4.8% 6.2% 8.6% 6.6%2013 6.5% 4.6% 6.0% 11.1% 5.9% 6.6% 4.8% 6.1% 8.8% 6.6%2014 6.5% 4.6% 6.0% 11.0% 5.9% 6.3% 4.8% 6.2% 9.1% 6.6%2015 6.3% 4.5% 5.9% 10.7% 5.8% 6.1% 4.7% 6.1% 8.4% 6.5%2016 6.3% 4.5% 5.8% 10.6% 5.8% 5.9% 4.8% 5.8% 8.3% 6.4%2017 6.3% 4.5% 5.8% 10.5% 5.8% 6.0% 4.8% 5.8% 8.2% 6.5%2018 6.3% 4.5% 5.8% 10.4% 5.8% 5.7% 4.8% 5.8% 8.2% 6.5%2019 6.2% 4.5% 5.8% 10.4% 5.8% 5.7% 4.8% 5.8% 8.3% 6.5%2020 6.2% 4.5% 5.8% 10.3% 5.8% 5.7% 4.8% 5.7% 8.1% 6.5%

SumYear\European CountryPassengers Growth

Historical

Forecast

260

Table C-3: Real 2000 GDP of U.S. and nine European countries during 2008 - 2020 (Source: United States Department of Agriculture

(USDA) International Macroeconomic Data Set).

Table C-4: Growth of Real 2000 GDP of U.S. and nine European countries during 2008 – 2020 (Source: United States Department of


0 1 2 3 4 5 6 7 8 9U.S. U.K. Germany France Netherlands Italy Ireland Spain Switzerland Belgium

2008 11,834,433 1,766,398 2,124,003 1,530,634 412,106 1,178,309 139,067 739,594 274,974 260,6732009 12,165,797 1,814,091 2,164,359 1,561,246 420,348 1,194,805 145,962 759,563 279,468 266,3262010 12,530,771 1,863,071 2,198,988 1,592,471 428,104 1,211,532 152,141 780,072 283,621 272,3022011 12,906,694 1,913,374 2,234,172 1,624,321 436,004 1,228,494 157,770 801,133 287,251 278,1842012 13,293,895 1,965,035 2,269,919 1,656,807 444,049 1,245,693 163,292 822,764 291,273 284,0532013 13,692,711 2,018,091 2,306,238 1,689,943 452,242 1,263,132 168,518 844,979 295,234 290,5012014 14,103,493 2,072,580 2,343,137 1,723,742 460,587 1,280,816 173,573 867,793 299,545 297,5312015 14,512,494 2,128,540 2,380,628 1,758,217 469,086 1,298,748 178,607 891,224 304,038 303,9582016 14,933,356 2,186,010 2,418,718 1,793,381 477,741 1,316,930 183,429 915,287 307,838 310,4322017 15,366,424 2,245,032 2,457,417 1,829,249 486,557 1,335,367 188,565 939,999 311,871 316,9512018 15,812,050 2,305,648 2,496,736 1,865,834 495,535 1,354,062 193,279 965,379 315,863 323,6392019 16,270,600 2,367,901 2,536,684 1,903,150 504,678 1,373,019 198,111 991,445 320,064 330,6942020 16,742,447 2,431,834 2,577,271 1,941,213 513,991 1,392,241 203,064 1,018,214 324,225 337,606

Year\European Country


2008 2.5% 2.7% 2.4% 2.1% 2.0% 1.6% 4.4% 2.8% 1.8% 2.0%2009 2.8% 2.7% 1.9% 2.0% 2.0% 1.4% 5.0% 2.7% 1.6% 2.2%2010 3.0% 2.7% 1.6% 2.0% 1.8% 1.4% 4.2% 2.7% 1.5% 2.2%2011 3.0% 2.7% 1.6% 2.0% 1.8% 1.4% 3.7% 2.7% 1.3% 2.2%2012 3.0% 2.7% 1.6% 2.0% 1.8% 1.4% 3.5% 2.7% 1.4% 2.1%2013 3.0% 2.7% 1.6% 2.0% 1.8% 1.4% 3.2% 2.7% 1.4% 2.3%2014 3.0% 2.7% 1.6% 2.0% 1.8% 1.4% 3.0% 2.7% 1.5% 2.4%2015 2.9% 2.7% 1.6% 2.0% 1.8% 1.4% 2.9% 2.7% 1.5% 2.2%2016 2.9% 2.7% 1.6% 2.0% 1.8% 1.4% 2.7% 2.7% 1.3% 2.1%2017 2.9% 2.7% 1.6% 2.0% 1.8% 1.4% 2.8% 2.7% 1.3% 2.1%2018 2.9% 2.7% 1.6% 2.0% 1.8% 1.4% 2.5% 2.7% 1.3% 2.1%2019 2.9% 2.7% 1.6% 2.0% 1.8% 1.4% 2.5% 2.7% 1.3% 2.2%2020 2.9% 2.7% 1.6% 2.0% 1.8% 1.4% 2.5% 2.7% 1.3% 2.1%


261

C.3 Comparison between Historical and Forecast Enplanements from the U.S. to Each of the Selected Nine

European Counties during 1990 – 2007

Figure C-11: Comparison between Historical and Forecast Enplanements from the U.S. to the U.K.

during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market Data.

Figure C-12: Comparison between Historical and Forecast Enplanements from the U.S. to Germany

during 1990 – 2007 (Historical Source: 1990 - 2007 T100 International Market Data).

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20074

4.5

5

5.5

6

6.5

7

7.5

8

8.5

9 x 106 1-United Kingdom (2007 Market Share: 33.44%) Yearly Passengers from U.S.

Year

Pass

enge

rs fr

om U

.S.


1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20072

2.5

3

3.5

4

4.5

5

5.5 x 106 2-Germany (2007 Market Share: 18.4%) Yearly Passengers from U.S. 1990 - 2007

Year

Pass

enge

rs fr

om U

.S.


262

Figure C-13: Comparison between Historical and Forecast Enplanements from the U.S. to France


Figure C-14: Comparison between Historical and Forecast Enplanements from the U.S. to Netherlands


1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20071.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8

3

3.2 x 106 3-France (2007 Market Share: 11.8%) Yearly Passengers from U.S. 1990 - 2007

Year

Pass

enge

rs fr

om U

.S.


1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20070.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6 x 106 4-Netherlands (2007 Market Share: 9%) Yearly Passengers from U.S. 1990 - 2007

Year

Pass

enge

rs fr

om U

.S.


263

Figure C-15: Comparison between Historical and Forecast Enplanements from the U.S. to Italy


Figure C-16: Comparison between Historical and Forecast Enplanements from the U.S. to Ireland


1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20070.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5 x 106 5-Italy (2007 Market Share: 5.7%) Yearly Passengers from U.S. 1990 - 2007

Year

Pass

enge

rs fr

om U

.S.


1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20072

3

4

5

6

7

8

9

10

11 x 105 6-Ireland (2007 Market Share: 4.2%) Yearly Passengers from U.S. 1990 - 2007

Year

Pass

enge

rs fr

om U

.S.


264

Figure C-17: Comparison between Historical and Forecast Enplanements from the U.S. to Spain


Figure C-18: Comparison between Historical and Forecast Enplanements from the U.S. to Switzerland


1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20074

5

6

7

8

9

10

11

12 x 105 7-Spain (2007 Market Share: 3.9%) Yearly Passengers from U.S. 1990 - 2007

Year

Pass

enge

rs fr

om U

.S.


1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20075

6

7

8

9

10

11 x 105 8-Switzerland (2007 Market Share: 3%) Yearly Passengers from U.S. 1990 - 2007

Year

Pass

enge

rs fr

om U

.S.


265

Figure C-19: Comparison between Historical and Forecast Enplanements from the U.S. to Belgium


1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 20072.5

3

3.5

4

4.5

5

5.5

6

6.5

7

7.5 x 105 9-Belgium (2007 Market Share: 1.6%) Yearly Passengers from U.S. 1990 - 2007

Year

Pass

enge

rs fr

om U

.S.


266

C.4 Input Data for the Model

Figure C-20: Enplanements from the United States to the Selected Nine European Countries in

Analysis Domain during 1990 - 2007 (Source: 1990 - 2007 T100 International Market Data).

‐

1

2

3

4

5

6

7

8

9

199019911992199319941995199619971998199920002001200220032004200520062007

Passen

gers from

U.S. (Millions)

Year

United Kingdom

Germany

France

Netherlands

Italy

Ireland

Spain

Switzerland

Belgium

267

Table C-5: Passenger Traffic between the United States and the Selected Nine European Countries in Analysis Domain during 1990 –

2007 (Source: 1990 – 2007 T100 International Market Data).

1 2 3 4 5 6 7 8 9

U.K. Germany France Netherlands Italy Ireland Spain Switzerland Belgium

1990 4,528,446 2,256,382 1,470,630 637,325 669,932 279,364 468,842 574,408 342,196 11,227,5251991 4,098,713 2,156,690 1,397,422 687,294 626,814 231,550 439,215 503,685 275,476 10,416,8591992 4,842,188 2,555,764 1,671,992 776,632 840,132 258,562 578,701 542,098 297,750 12,363,8191993 5,174,786 2,657,348 1,606,463 964,422 822,010 248,994 519,705 581,136 335,986 12,910,8501994 5,333,169 2,690,340 1,738,517 1,094,743 882,644 316,378 577,783 628,521 345,132 13,607,2271995 5,759,648 2,845,646 1,765,418 1,186,835 920,484 338,800 605,000 680,574 366,790 14,469,1951996 6,141,315 2,912,320 1,949,248 1,404,848 953,423 416,048 567,609 729,683 389,560 15,464,0541997 6,829,973 3,151,366 2,056,891 1,652,792 1,023,953 414,485 619,862 774,321 492,387 17,016,0301998 7,606,402 3,247,833 2,208,808 2,000,085 1,009,600 484,939 690,228 885,156 613,387 18,746,4381999 8,206,666 3,504,471 2,369,384 2,189,654 1,148,349 630,692 767,081 990,856 695,452 20,502,6052000 8,665,584 3,723,927 3,008,111 2,265,435 1,461,210 748,719 800,155 1,050,556 728,523 22,452,2202001 7,660,202 3,338,407 2,878,633 2,017,044 1,202,177 743,481 682,384 909,585 575,957 20,007,8702002 7,692,186 3,277,681 2,706,402 1,974,929 1,001,086 639,532 653,740 717,803 260,257 18,923,6162003 7,552,207 3,403,578 2,556,152 1,959,699 950,437 774,089 653,147 698,051 261,028 18,808,3882004 8,216,281 3,807,120 2,779,885 2,139,464 1,205,983 829,352 765,522 704,368 331,242 20,779,2172005 8,344,689 4,093,115 2,986,896 2,272,816 1,277,667 909,869 811,046 699,668 363,961 21,759,7272006 8,276,607 4,285,164 2,973,593 2,238,902 1,314,422 996,896 858,197 720,786 360,237 22,024,8042007 8,513,976 4,677,240 3,013,163 2,344,010 1,443,529 1,074,759 1,003,373 794,974 396,021 23,261,045

Year \ European Country

Sum

268

Table C-6: Growth of Passenger Traffic between the United States and the Selected Nine European Countries in Analysis Domain

during 1990 – 2007 (Source: 1990 – 2007 T100 International Market Data).

1 2 3 4 5 6 7 8 9

U.K. Germany France Netherlands Italy Ireland Spain Switzerland Belgium

1990 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐1991 ‐9.5% ‐4.4% ‐5.0% 7.8% ‐6.4% ‐17.1% ‐6.3% ‐12.3% ‐19.5% ‐7.2%1992 18.1% 18.5% 19.6% 13.0% 34.0% 11.7% 31.8% 7.6% 8.1% 18.7%1993 6.9% 4.0% ‐3.9% 24.2% ‐2.2% ‐3.7% ‐10.2% 7.2% 12.8% 4.4%1994 3.1% 1.2% 8.2% 13.5% 7.4% 27.1% 11.2% 8.2% 2.7% 5.4%1995 8.0% 5.8% 1.5% 8.4% 4.3% 7.1% 4.7% 8.3% 6.3% 6.3%1996 6.6% 2.3% 10.4% 18.4% 3.6% 22.8% ‐6.2% 7.2% 6.2% 6.9%1997 11.2% 8.2% 5.5% 17.6% 7.4% ‐0.4% 9.2% 6.1% 26.4% 10.0%1998 11.4% 3.1% 7.4% 21.0% ‐1.4% 17.0% 11.4% 14.3% 24.6% 10.2%1999 7.9% 7.9% 7.3% 9.5% 13.7% 30.1% 11.1% 11.9% 13.4% 9.4%2000 5.6% 6.3% 27.0% 3.5% 27.2% 18.7% 4.3% 6.0% 4.8% 9.5%2001 ‐11.6% ‐10.4% ‐4.3% ‐11.0% ‐17.7% ‐0.7% ‐14.7% ‐13.4% ‐20.9% ‐10.9%2002 0.4% ‐1.8% ‐6.0% ‐2.1% ‐16.7% ‐14.0% ‐4.2% ‐21.1% ‐54.8% ‐5.4%2003 ‐1.8% 3.8% ‐5.6% ‐0.8% ‐5.1% 21.0% ‐0.1% ‐2.8% 0.3% ‐0.6%2004 8.8% 11.9% 8.8% 9.2% 26.9% 7.1% 17.2% 0.9% 26.9% 10.5%2005 1.6% 7.5% 7.4% 6.2% 5.9% 9.7% 5.9% ‐0.7% 9.9% 4.7%2006 ‐0.8% 4.7% ‐0.4% ‐1.5% 2.9% 9.6% 5.8% 3.0% ‐1.0% 1.2%2007 2.9% 9.1% 1.3% 4.7% 9.8% 7.8% 16.9% 10.3% 9.9% 5.6%

Year \ European Country

Sum

269

Table C-7: Real 2000 GDP of U.S. and Nine European Countries during 1990 – 2007 (Source: United States Department of Agriculture

(USDA) International Macroeconomic Data Set).


1990 7,112,500 1,136,040 1,586,630 1,089,006 276,583 918,137 47,165 434,608 220,373 183,7101991 7,100,500 1,120,617 1,625,029 1,100,599 282,859 930,901 48,075 444,466 218,608 187,2561992 7,336,600 1,123,136 1,655,319 1,114,532 288,589 937,981 49,682 447,520 218,332 190,2301993 7,532,700 1,148,766 1,642,215 1,105,494 290,786 929,691 51,020 442,313 217,281 187,3371994 7,835,500 1,198,226 1,686,891 1,129,983 300,159 950,214 53,956 452,270 218,438 192,9091995 8,031,700 1,233,329 1,720,124 1,155,761 308,520 977,328 59,300 465,150 222,655 199,4881996 8,328,900 1,267,278 1,737,121 1,168,552 317,897 988,011 64,088 476,486 223,817 201,8411997 8,703,500 1,305,773 1,769,405 1,194,542 330,099 1,008,031 71,196 495,671 228,087 208,8101998 9,066,900 1,349,437 1,801,927 1,237,052 344,457 1,026,115 77,349 517,210 234,457 213,0261999 9,470,300 1,390,218 1,835,825 1,276,856 358,220 1,043,186 86,076 539,056 237,537 219,8472000 9,817,000 1,443,190 1,899,301 1,328,494 370,638 1,074,763 94,956 580,673 246,049 228,4172001 9,890,700 1,477,138 1,925,223 1,352,017 375,927 1,093,724 100,661 601,255 248,612 230,0592002 10,048,800 1,507,556 1,925,318 1,366,737 378,064 1,097,917 106,835 617,369 249,421 232,1382003 10,301,000 1,547,741 1,921,473 1,381,362 374,743 1,100,708 110,738 635,360 248,541 235,0612004 10,676,000 1,598,241 1,936,285 1,412,523 380,136 1,114,184 116,133 655,007 253,761 241,9102005 11,003,000 1,629,119 1,957,507 1,436,812 385,951 1,115,377 121,541 677,451 258,454 245,6332006 11,319,400 1,674,125 2,015,951 1,468,362 395,540 1,137,005 127,375 699,174 265,648 250,9042007 11,545,788 1,719,959 2,075,094 1,499,151 404,025 1,159,745 133,208 719,450 270,127 255,523

Year\U.S. + European Country

270

Table C-8: Growth of Real 2000 GDP of U.S. and Nine European Countries during 1990 – 2007 (Source: United States Department of



1990 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐1991 ‐0.2% ‐1.4% 2.4% 1.1% 2.3% 1.4% 1.9% 2.3% ‐0.8% 1.9%1992 3.3% 0.2% 1.9% 1.3% 2.0% 0.8% 3.3% 0.7% ‐0.1% 1.6%1993 2.7% 2.3% ‐0.8% ‐0.8% 0.8% ‐0.9% 2.7% ‐1.2% ‐0.5% ‐1.5%1994 4.0% 4.3% 2.7% 2.2% 3.2% 2.2% 5.8% 2.3% 0.5% 3.0%1995 2.5% 2.9% 2.0% 2.3% 2.8% 2.9% 9.9% 2.8% 1.9% 3.4%1996 3.7% 2.8% 1.0% 1.1% 3.0% 1.1% 8.1% 2.4% 0.5% 1.2%1997 4.5% 3.0% 1.9% 2.2% 3.8% 2.0% 11.1% 4.0% 1.9% 3.5%1998 4.2% 3.3% 1.8% 3.6% 4.3% 1.8% 8.6% 4.3% 2.8% 2.0%1999 4.4% 3.0% 1.9% 3.2% 4.0% 1.7% 11.3% 4.2% 1.3% 3.2%2000 3.7% 3.8% 3.5% 4.0% 3.5% 3.0% 10.3% 7.7% 3.6% 3.9%2001 0.8% 2.4% 1.4% 1.8% 1.4% 1.8% 6.0% 3.5% 1.0% 0.7%2002 1.6% 2.1% 0.0% 1.1% 0.6% 0.4% 6.1% 2.7% 0.3% 0.9%2003 2.5% 2.7% ‐0.2% 1.1% ‐0.9% 0.3% 3.7% 2.9% ‐0.4% 1.3%2004 3.6% 3.3% 0.8% 2.3% 1.4% 1.2% 4.9% 3.1% 2.1% 2.9%2005 3.1% 1.9% 1.1% 1.7% 1.5% 0.1% 4.7% 3.4% 1.8% 1.5%2006 2.9% 2.8% 3.0% 2.2% 2.5% 1.9% 4.8% 3.2% 2.8% 2.1%2007 2.0% 2.7% 2.9% 2.1% 2.1% 2.0% 4.6% 2.9% 1.7% 1.8%


271

Table C-9: Population of U.S. and Nine European Countries during 1990 – 2007 (Source: United States Department of Agriculture (USDA)

International Macroeconomic Data Set).


1990 250,131,894 57,493,307 79,380,394 56,735,161 14,951,510 56,742,886 3,508,200 39,350,769 6,836,626 9,969,310 1991 253,492,503 57,665,646 79,984,244 57,055,448 15,066,220 56,747,462 3,530,771 39,461,418 6,920,562 10,004,4871992 256,894,189 57,866,349 80,597,764 57,374,179 15,174,244 56,840,847 3,557,761 39,549,438 6,995,447 10,045,6221993 260,255,352 58,026,920 81,132,272 57,658,289 15,274,942 57,026,746 3,578,349 39,627,587 7,058,211 10,085,4261994 263,435,673 58,212,518 81,414,164 57,906,847 15,382,198 57,179,460 3,595,542 39,690,971 7,114,530 10,122,9141995 266,557,091 58,426,014 81,653,702 58,149,727 15,459,054 57,274,531 3,613,890 39,749,715 7,157,106 10,155,4591996 269,667,391 58,618,663 81,890,667 58,388,408 15,527,809 57,367,032 3,636,179 39,803,829 7,181,024 10,178,9341997 272,911,760 58,808,266 82,011,073 58,623,428 15,604,464 57,479,469 3,667,233 39,855,442 7,193,761 10,199,7871998 276,115,288 59,035,652 82,023,672 58,866,290 15,699,259 57,550,318 3,707,555 39,906,235 7,207,995 10,217,0301999 279,294,713 59,293,320 82,074,778 59,116,128 15,801,947 57,603,634 3,750,141 39,953,263 7,232,809 10,235,6552000 282,338,631 59,522,468 82,187,909 59,381,628 15,907,853 57,719,337 3,791,690 40,016,081 7,266,920 10,263,6182001 285,023,886 59,723,243 82,280,551 59,658,144 16,017,445 57,844,924 3,835,025 40,087,104 7,311,237 10,291,6792002 287,675,526 59,912,431 82,350,671 59,925,035 16,122,830 57,926,999 3,879,155 40,152,517 7,361,757 10,311,9702003 290,342,554 60,094,648 82,398,326 60,180,529 16,223,248 57,998,353 3,924,023 40,217,413 7,408,319 10,330,8242004 293,027,571 60,270,708 82,424,609 60,424,213 16,318,199 58,057,477 3,969,558 40,280,780 7,450,867 10,348,2762005 295,734,134 60,441,457 82,431,390 60,656,178 16,407,491 58,103,033 4,015,676 40,341,462 7,489,370 10,364,3882006 298,444,215 60,609,153 82,422,299 60,876,136 16,491,461 58,133,509 4,062,235 40,397,842 7,523,934 10,379,0672007 301,139,947 60,776,238 82,400,996 61,083,916 16,570,613 58,147,733 4,109,086 40,448,191 7,554,661 10,392,226


272

Table C-10: Average Airfare (2000 Year $) from U.S. to Nine European Countries during 1998 – 2007 (Source: 1998 – 2007 DB1B Data).


1998 605 617 493 571 520 500 481 675 601 1999 545 576 445 497 477 434 459 606 561 2000 569 532 462 493 461 464 462 556 553 2001 488 501 472 493 457 438 462 546 496 2002 467 548 496 537 491 407 475 592 572 2003 462 539 477 504 454 383 446 550 553 2004 483 553 500 528 479 416 460 553 570 2005 513 585 524 557 508 388 485 551 599 2006 518 578 542 575 547 377 520 578 600 2007 543 578 576 605 588 399 527 588 619


273

Appendix D:

Domestic Leg of International Passengers within the Continental U.S. (CONUS)

274

D.1 Summary of Matlab Functions

DB1A_DB1B_Convert.m % This function converts the DB1A/DB1B raw data into Ticket, Coupon and % Market data, respectively, by itinerary geographic type % 0/1 - International/non-continental % 2 - Domestic(continental) % (The raw data dictionary is based on the document titled % 'BTS Database DB1A Record Generation (04/16/2003)' % % Calling: % N/A % % Called: % Main_DB1A_DB1B_Convert Input DB1A_Y1990_Q1 (Predefined input)

DB1A_Y1990_Q2 (Predefined input) DB1A_Y1990_Q3 (Predefined input) DB1A_Y1990_Q4 (Predefined input) . DB1B_Y1998_Q1 (Predefined input) . DB1B_Y2007_Q4 (Predefined input)

Output DB1A.I.Ticket_Y1990_Q1DB1A.I.Ticket_Y1990_Q2DB1A.I.Ticket_Y1990_Q3DB1A.I.Ticket_Y1990_Q4. DB1A.I.Ticket_Y1997_Q4DB1B.I.Ticket_Y1998_Q1. DB1B.I.Ticket_Y2007_Q4

DB1A.I.Coupon_Y1990_Q1DB1A.I.Coupon_Y1990_Q2DB1A.I.Coupon_Y1990_Q3DB1A.I.Coupon_Y1990_Q4. DB1A.I.Coupon_Y1997_Q4DB1B.I.Coupon_Y1998_Q1. DB1B.I.Coupon_Y2007_Q4

DB1A.I.Market_Y1990_Q1DB1A.I.Market_Y1990_Q2DB1A.I.Market_Y1990_Q3DB1A.I.Market_Y1990_Q4. DB1A.I.Market_Y1997_Q4DB1B.I.Market_Y1998_Q1. DB1B.I.Market_Y2007_Q4

DB1A.D.Ticket_Y1990_Q1DB1A.D.Ticket_Y1990_Q2DB1A.D.Ticket_Y1990_Q3DB1A.D.Ticket_Y1990_Q4. DB1A.D.Ticket_Y1997_Q4DB1B.D.Ticket_Y1998_Q1. DB1B.D.Ticket_Y2007_Q4

DB1A.D.Coupon_Y1990_Q1DB1A.D.Coupon_Y1990_Q2DB1A.D.Coupon_Y1990_Q3DB1A.D.Coupon_Y1990_Q4. DB1A.D.Coupon_Y1997_Q4DB1B.D.Coupon_Y1998_Q1. DB1B.D.Coupon_Y2007_Q4

DB1A.D.Market_Y1990_Q1DB1A.D.Market_Y1990_Q2DB1A.D.Market_Y1990_Q3DB1A.D.Market_Y1990_Q4. DB1A.D.Market_Y1997_Q4DB1B.D.Market_Y1998_Q1. DB1B.D.Market_Y2007_Q4

275

Main_Historical_1.m % This function identifies the U.S. carriers operated DB1B sampled % international and non-CONUS itineraries and the gateway airport for each % of the itinerary. The scale_factors are then estimated to adjust the % sampled international and non-CONUS itineraries to 100% level. The DOI % enplanements between TSAM airport is finally estimated based on the % adjusted 100% itineraries. % Calling: % Identify_TSAM_Airport_ID_OAG % DB1B_I_Identify_Itin_US_Carriers % DB1B_I_Estimate_Dist_No_of_Domestic_Legs_US_Carriers % Estimate_TSAM_DOI_by_Coupon_Scaled % % Called: % Main Input DB1A_I_Coupon_Dir (Predefined variable)

DB1B_I_Coupon_Dir (Predefined variable) TSAM_Airport_List.mat (Predefined input) Carrier_Group_Revised.txt (Predefined input) T100_MARKET_International_Enplanements by Airport_1990-2008.txt (Predefined input) T100_MARKET_International_Enplanements by Airport_1990-2008_US.txt (Predefined input) T100_MARKET_US Territories_Enplanements by Airport_1990-2008.txt (Predefined input) Threshold_Gateway_Enpls_DB1B (Predefined variable) Threshold_Gateway_Enpls_T100 (Predefined variable)

Output Itin_US_Carriers.mat Gateway_Summary.mat DB1B_I_Dist_No_of_Domestic_Legs_US_Carrier.mat TSAM_DOI_by_Coupon_Scaled.mat DB1B_I_Not_in_TSAM_Scaled.mat

DB1B_I_Identify_Itin_US_Carriers.m% This function identifies all the U.S. carriers operated international and % non-CONUS itineraries sampled in DB1B data and the U.S. gateway airport. % It also estimated the scale_factor_1 as the ratio of the DB1B sampled % U.S. carriers operated international and non-CONUS enplanements over the % corresponding T100 international and non-CONUS enplanements, % respectively, and the scale_factor_2 as the ratio of the T100 U.S. % carriers operated international enplanements over the total international % enplanements % % Calling: % Read_DB1B_Coupon % % Called: % Main_Historical_1 Input TSAM_Airport_List.mat (Predefined input)

Carrier_Group_Revised.txt (Predefined input) T100_MARKET_International_Enplanements by Airport_1990-2008.txt (Predefined input) T100_MARKET_International_Enplanements by Airport_1990-2008_US.txt (Predefined input) T100_MARKET_US Territories_Enplanements by Airport_1990-2008.txt (Predefined input)

Output Itin_US_Carriers.mat Gateway_Summary.mat

276

DB1B_I_Estimate_Dist_No_of_Domestic_Legs_US_Carriers.m% This function estimates the distribution of the no. of the domestic leg % of U.S. carriers operated international and non-CONUS itineraries sampled % in the DB1B data % % Calling: % Read_DB1B_Coupon % % Called: % Main_Historical_1 Input DB1A_I_Coupon_Dir (Predefined variable)

DB1B_I_Coupon_Dir (Predefined variable) Itin_US_Carriers.mat

Output DB1B_I_Dist_No_of_Domestic_Legs_US_Carrier.mat Identify_TSAM_Airport_ID_OAG.m% This function identifies TSAM airport set for each analyze year % % Calling: % N/A % % Called: % Main_Historical_1 % Main_Historical_2 % Summarize_TSAM_DOI_by_Airport_Scaled_Time_SeriesInput Year

TSAM_Airport_List.mat (Predefined input) Output TSAM_Airport_ID_OAG.mat Estimate_TSAM_DOI_by_Coupon_Scaled.m% This function estimates the number of DOI between TSAM airports by % adjusting the sampled U.S. carriers operated DB1B international and % non-CONUS itineraries by the scale factors estimated for each gateway % airport % % Calling: % Read_DB1B_Coupon % % Called: % Main_Historical_1 Input DB1A_I_Coupon_Dir (Predefined variable)

DB1B_I_Coupon_Dir (Predefined variable) TSAM_Airport_ID_OAG.mat Threshold_Gateway_Enpls_DB1B (Predefined variable) Threshold_Gateway_Enpls_T100 (Predefined variable) Itin_US_Carriers.mat Gateway_Summary.mat

Output TSAM_DOI_by_Coupon_Scaled.mat DB1B_I_Not_in_TSAM_Scaled.mat

277

Main_Historical_2.m % This function summarizes the number of DOI enplanements by flight coupon, % airport and total number of DOI enplanements based on the adjusted DB1B % sampled international and non-CONUS itineraries % Calling: % Identify_TSAM_Airport_ID_OAG % Summarize_TSAM_DOI_by_Coupon_Scaled_Sorted % Summarize_TSAM_DOI_by_Coupon_Scaled_Time_Series % Summarize_TSAM_DOI_by_Airport_Scaled % Summarize_TSAM_DOI_by_Airport_Scaled_Time_Series % % Called: % Main Input Years_Analyzed (Predefined variable)

TSAM_Airport_List.mat (Predefined input)Output TSAM_DOI_Total_Scaled.mat

TSAM_DOI_by_Coupon_Scaled_Sorted.mat TSAM_DOI_by_Coupon_Scaled_1990_2007.mat TSAM_DOI_by_Airport_Scaled.mat TSAM_DOI_by_Airport_Scaled_1990_2007.mat

Summarize_TSAM_DOI_by_Coupon_Scaled_Sorted.m% This function summarizes the DOI enplanements by flight coupon in % decreasing order for the historical years (1990-2007) or the forecast % years (2008-2040) % % Calling: % N/A % % Called: % Main_Historical_2 % Main_Forecast Input TSAM_Airport_ID_OAG.mat

TSAM_DOI_by_Coupon_Scaled.matOutput TSAM_DOI_by_Coupon_Scaled_Sorted.mat

Summarize_TSAM_DOI_by_Airport_Scaled.m% This function summarizes the DOI enplanements by airport in decreasing % order for the historical years (1990-2007) % % Calling: % N/A % % Called: % Main_Historical_2 Input TSAM_Airport_ID_OAG.mat

TSAM_DOI_by_Coupon_Scaled.matOutput TSAM_DOI_by_Airport_Scaled.mat

278

Summarize_TSAM_DOI_by_Coupon_Scaled_Time_Series.m% This function summarizes the timer series of scaled number of DOI % enplanements by flight coupon during the historical years (1990-2007) or % the forecast years (2008-2040) % % Calling: % N/A % % Called: % Main_Historical_2 % Main_Forecast Input Years_Analyzed (Predefined variable)

TSAM_DOI_by_Coupon_Scaled_Sorted.matOutput TSAM_DOI_by_Coupon_Scaled_1990_2007.mat

Summarize_TSAM_DOI_by_Airport_Scaled_Time_Series.m% This function summarizes the DOI enplanements by airport in decreasing % order for the historical years (1990-2007) or the forecast years % (2008-2040) % % Calling: % N/A % % Called: % Main_Historical_2 % Main_Forecast Input Years_Analyzed (Predefined variable)

TSAM_Airport_List.mat (Predefined input) TSAM_DOI_by_Airport_Scaled.mat

Output TSAM_DOI_by_Airport_Scaled_1990_2007.mat

Main_Forecast.m % This function forecasts the total number of domestic enplanements created % by the international passengers, and then estimate the total international % passengers on each domestic leg for all the future years using Fratar % model % % Calling: % Forecast_TSAM_DOI_Total_Scaled % Identify_TSAM_Airport_ID_OAG % Forecast_TSAM_DOI_by_Airport_Scaled % Fratar_Model % Update_TSAM_DOI_by_Coupon_Scaled_Forecast % Summarize_TSAM_DOI_by_Coupon_Scaled_Sorted % Summarize_TSAM_DOI_by_Coupon_Scaled_Time_Series % Summarize_TSAM_DOI_by_Airport_Scaled_Time_Series % % Called: % Main Input Historical_Years_Analyzed (Predefined variable)

Forecast_Years_Analyzed (Predefined variable) T100_US_to_WorldRegion_Pax_1990_2007.mat (Predefined input) T100_US_to_WorldRegion_Pax_2008_2040.mat (Predefined input) TSAM_DOI_Total_Scaled.mat

Output TSAM_DOI_Total_Scaled_Forecast.mat TSAM_DOI_by_Coupon_Scaled_Forecast.mat TSAM_DOI_by_Coupon_Scaled_Sorted_Forecast.mat TSAM_DOI_by_Airport_Scaled_Forecast.mat TSAM_DOI_by_Coupon_Scaled_2008_2040.mat

279

TSAM_DOI_by_Airport_Scaled_2008_2040.mat Forecast_TSAM_DOI_Total_Scaled.m% This function forecasts the total number of DOI enplanements for the % future years % % Calling: % regress (Matlab built-in tool box) % % Called: % Main_Forecast Input Historical_Years_Analyzed (Predefined variable)

Forecast_Years_Analyzed (Predefined variable) T100_US_to_WorldRegion_Pax_1990_2007.mat (Predefined input) T100_US_to_WorldRegion_Pax_2008_2040.mat (Predefined input) TSAM_DOI_Total_Scaled.mat

Output TSAM_DOI_Total_Scaled_Forecast.mat

Forecast_TSAM_DOI_by_Airport_Scaled.m% This function obtains the forecast DOI enplanements/deplanements at the % airport for the future years (2008-2040) % Calling: % N/A % % Called: % Main_Forecast Input TSAM_Airport_ID_OAG.mat

TSAM_DOI_Total_Scaled_Forecast.mat TSAM_DOI_Total_Scaled_Base_Year.mat TSAM_DOI_by_Airport_Scaled_Base_Year.mat TSAM_Airport_ID_OAG_Base_Year.mat

Output TSAM_DOI_by_Airport_Scaled_Forecast.mat

Fratar_Model.m % This function obtains the forecast DOI enplanements between TSAM airports % from the forecast DOI enplanements between base year (2007) TSAM airports % % Calling: % N/A % % Called: % Main_Forecast Input TSAM_DOI_by_Coupon_Scaled_Base_Year.mat

Growth_Factor.mat TSAM_DOI_by_Airport_Scaled_Base_Year.mat Alpha (Predefined variable) Max_Iteration (Predefined variable)

Output TSAM_DOI_by_Coupon_Scaled_Forecast.mat

Update_TSAM_DOI_by_Coupon_Scaled_Forecast.m% This function obtains the forecast DOI enplanements between TSAM airports % from the forecast DOI enplanements between base year (2007) TSAM airports % % Calling: % N/A % % Called: % Main_Forecast

280

Input TSAM_DOI_by_Coupon_Scaled_Forecast.matOutput TSAM_DOI_by_Coupon_Scaled_Forecast.mat

Summarize_TSAM_DOI_by_Coupon_Scaled_Sorted.m% This function summarizes the DOI enplanements by flight coupon in % decreasing order for the historical years (1990-2007) or the forecast % years (2008-2040) % % Calling: % N/A % % Called: % Main_Historical_2 % Main_Forecast Input TSAM_Airport_ID_OAG.mat

TSAM_DOI_by_Coupon_Scaled.matOutput TSAM_DOI_by_Coupon_Scaled_Sorted_Forecast.mat

Summarize_TSAM_DOI_by_Coupon_Scaled_Time_Series.m% This function summarizes the timer series of scaled number of DOI % enplanements by flight coupon during the historical years (1990-2007) or % the forecast years (2008-2040) % % Calling: % N/A % % Called: % Main_Historical_2 % Main_Forecast Input Years_Analyzed (Predefined variable)

TSAM_DOI_by_Coupon_Scaled_Sorted.matOutput TSAM_DOI_by_Coupon_Scaled_2008_2040.mat

Summarize_TSAM_DOI_by_Airport_Scaled_Time_Series.m% This function summarizes the DOI enplanements by airport in decreasing % order for the historical years (1990-2007) or the forecast years % (2008-2040) % % Calling: % N/A % % Called: % Main_Historical_2 % Main_Forecast Input Years_Analyzed (Predefined variable)

TSAM_Airport_List.mat (Predefined variable) TSAM_DOI_by_Airport_Scaled.mat

Output TSAM_DOI_by_Airport_Scaled_2008_2040.mat

281

D.2 Description of Input and Output Variables for Matlab Functions

Input/Output Variable Dimension of Variable

Remark on Dimension of Variable

Dimension of Cell

Remark on Dimension of Cell

Gateway_Summary_1990 . . . Gateway_Summary_2007

2 x 1 Itineraty_Geo_Type (1 – International; 2 – Non‐CONUS) x Cell

No. of gateways x 5

1 – Gateway airport 2 – International or non‐CONUS enplanements by U.S. carriers from T100 3 – International or non‐CONUS enplanements by U.S. carriers sampled in DB1B 4 – Scale factor 1 (Field 2/ Field 3) 5 – Scale factor 2 (Field 2/ International or non‐CONUS enplanements by both U.S. carriers and foreign carriers from T100)

Itin_US_Carriers_1990 . . . Itin_US_Carriers_2007

4 x 1 Quarter x Cell No. of itineraries x 3

1 ‐ Itinerary_ID 2 – Gateway index 3 – Itinerary_Geo_Type (1 – International; 2 – Non‐CONUS)

TSAM_DOI_Total_Scaled 18 x 2 Year (1990‐2007) x Total DOI

‐ ‐

TSAM_DOI_Total_Scaled_Forecast 33 x 2 Year (2008‐2040) x Total DOI

‐ ‐

TSAM_DOI_by_Coupon_Scaled_1990 . . . TSAM_DOI_by_Coupon_Scaled_2007

2 x 1 Itineraty_Geo_Type (1 – International; 2 – Non‐CONUS) x Cell

533 x 533 . . . 417 x 417

Airport x Airport

TSAM_DOI_by_Coupon_Scaled_Forecast 33 x 2 Year (2008‐2040) x Cell

378 x 378 Airport x Airport

282

TSAM_DOI_by_Coupon_Scaled_Sorted 18 x 2 Year (1990‐2007) x Cell

No. of coupons (4569, 4635, …, 4989) x 3

1 – Coupon 2 – Direction as specified 3 – Direction reversed

TSAM_DOI_by_Coupon_Scaled_Sorted_Forecast 33 x 2 Year (2008‐2040) x Cell

4936 x 3 1 – Coupon 2 – Direction as specified 3 – Direction reversed

TSAM_DOI_by_Airport_Scaled 18 x 2 Year (1990‐2007) x Cell

No. of airports (533, 451, 406, 417) x 3

1 – Airport 2 – DOI Enplanements 3 – DOI Deplanements

TSAM_DOI_by_Airport_Scaled_Forecast 33 x 2 Year (2008‐2040) x Cell

378 x 3 1 – Airport 2 – DOI Enplanements 3 – DOI Deplanements

TSAM_DOI_by_Coupon_Scaled_1990_2007 13690 x 19Coupon x Year (1990‐2007)

‐ ‐

TSAM_DOI_by_Coupon_Scaled_2008_2040 4936 x 34 Coupon x Year (2008‐2040)

‐ ‐

TSAM_DOI_by_Airport_Scaled_1990_2007 546 x 19 Airport x Year (1990‐2007)

‐ ‐

TSAM_DOI_by_Airport_Scaled_2008_2040 378 x 34 Airport x Year (2008‐2040)

‐ ‐

283

Appendix E:

Development of Secondary Airports in Multi-Airport Systems (MAS) in the NAS

284

E.1 Evolution of Enplanements Share at Airports in Multi-Airport System

Figure E-1: Evolution of Enplanements Share at Airports in the New York City MAS


Figure E-2: Evolution of Enplanements Share at Airports in the Chicago MAS


0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

Enplan

emen

ts Sha

re in

MAS of

New

York

JFK (New York/Kenndy)EWR (New York/Newark)LGA (New York/LaGuardia)ISP (New York/Islip)HPN (New York/Westchester)SWF (New York/Stewart)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Enplan

emen

ts Sha

re in

MAS of

Chicago

ORD (Chicago/O'Hare)

MDW (Chicago/Midway)

285

Figure E-3: Evolution of Enplanements Share at Airports in the Los Angeles MAS


Figure E-4: Evolution of Enplanements Share at Airports in the Dallas MAS


Figure E-5: Evolution of Enplanements Share at Airports in the Washington D.C. MAS


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

Enplan

emen

ts Sha

re in

MAS of

Los A

ngeles

LAX (Los Angeles/Int'l)

SNA (Los Angeles/Orange County)

ONT (Los Angeles/Ontario)

BUR (Los Angeles/Burbank)

LGB (Los Angeles/Long Beach)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Enplan

emen

ts Sha

re in

MAS of

Dallas

DFW (Dallas/Fort Worth)

DAL (Dallas/Love Field)

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

Enplan

emen

ts Sha

re in

MAS of

Washinton

D.C.

IAD (Washington/Dulles)

BWI (Washington/Baltimore)

DCA (Wasington/Reagan)

286

Figure E-6: Evolution of Enplanements Share at Airports in the San Francesco MAS


Figure E-7: Evolution of Enplanements Share at Airports in the Orlando MAS


Figure E-8: Evolution of Enplanements Share at Airports in the Detroit MAS


0%

10%

20%

30%

40%

50%

60%

70%

80%

Enplan

emen

ts Sha

re in

MAS of

San Fran

cesco SFO (San Francesco/Int'l)

OAK (San Francesco/Oakland)

SJC (San Francesco/San Jose)

0%

20%

40%

60%

80%

100%

120%

Enplan

emen

ts Sha

re in

MAS of

Orlan

do

MCO (Orlando/Int'l)

SFB (Orlando/Sanford)

0%

20%

40%

60%

80%

100%

120%

Enplan

emen

ts Sha

re in

MAS of

Detroit DTW (Detroit/Metropolitan)

FNT (Detroit/Bishop)

287

Figure E-9: Evolution of Enplanements Share at Airports in the Philadelphia MAS


Figure E-10: Evolution of Enplanements Share at Airports in the Tampa MAS


Figure E-11: Evolution of Enplanements Share at Airports in the Cincinnati MAS


0%

20%

40%

60%

80%

100%

120%

Enplan

emen

ts Sha

re in

MAS of

Philade

lphia

PHL (Philadelphia/Int'l)

ACY (Philadelphia/Atlantic City)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Enplan

emen

ts Sha

re in

MAS of

Tampa TPA (Tampa/Int'l)

SRQ (Tampa/Sarasota)

PIE (Tampa/St. Petersburg)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Enplan

emen

ts Sha

re in

MAS of

Cincinna

ti

CVG (Cincinnati/Int'l)

DAY (Cincinnati/Dayton)

288

Figure E-12: Evolution of Enplanements Share at Airports in the Cleveland MAS


Figure E-13: Evolution of Enplanements at Airports in the Houston MAS


0%

20%

40%

60%

80%

100%

120%

Enplan

emen

ts Sha

re in

MAS of

Clevelan

d CLE (Cleveland/Hopkins)


‐

5

10

15

20

25

Enplan

emen

ts

Total at IAHLCCs' at IAHTotal at HOULCCs' at HOU

289

E.2 Non-Stop Markets Served at Airports in Multi-Airport System & Traffic Distribution by

Market/Carrier among Airports in Multi-Airport System

Figure E-14: Non-Stop Markets Served in the New York City MAS – JFK* (Kennedy),

EWR* (Newark), LGA* (LarGuardia), ISP (Islip), HPN (Westchester), SWF (Stewart)

(The airport marked by * is the primary airport. This applys to the rest of this section).

JFK (Year = 2008) EWR (Year = 2008)

LGA (Year = 2008) ISP (Year = 2008)

HPN (Year = 2008) SWF (Year = 2008)

290

Figure E-15: Non-Stop Markets Served in the Chicago MAS.

Figure E-16: Traffic Distribution by Market/Carrier among Airports in the Chicago MAS.

ORD (Year = 2008)

MDW (Year = 2008)

0.0 2.0 4.0

LGA,EWR,JFK,HPN,I…DCA,BWI,IADLAX,SNA,LGBBOS,PVD,MHTSFO,SJC,OAK

MSPATL

DTW,FNTPHL,ABE

DENSTL

MIA,FLL,PBILAS

DFW,DALPHXMCIMCO

IAH,HOUSEA

CLE,CAKCMH

TPA,SRQBNASANCLT

Seats (Millions)

ORD

MDW

02468

101214161820

Seats C

apacity (M

illions)

MDW (Chicago/Midway)ORD (Chicago/O'Hare)

291

Figure E-17: Non-Stop Markets Served in the Los Angeles MAS – LAX* (Int’l), SNA (Orange

County), ONT (Ontario), BUR (Burbank), LGB (Long Beach).

LAX (Year = 2008)

SNA (Year = 2008) ONT (Year = 2008)

BUR (Year = 2008) LGB (Year = 2008)

292

Figure E-18: Non-Stop Markets Served in the Dallas MAS

Figure E-19: Traffic Distribution by Market/Carrier among Airports in the Dallas MAS.

DFW (Year = 2008)

DAL (Year = 2008)

0.0 1.0 2.0 3.0

HOU,IAHSATAUSORDSTLABQMCIELPTULMSYOKCLBBAMALIT

MAFBHM

Seats (Millions)

Non

‐stop Markets

DALDFW

0

5

10

15

20

25

30

Seats C

apacity (M

illions)

DAL (Dallas/Love Field)

DFW (Dallas/Fort Worth)

293

Figure E-20: Non-Stop Markets Served in the Houston MAS

Figure E-21: Traffic Distribution by Market/Carrier among Airports in Houston MAS.

IAH (Year = 2008)

HOU (Year = 2008)

0.0 0.5 1.0 1.5 2.0 2.5

DAL,DFWLAX,SNA,ONTEWR,LGA,JFK

MSYORD,MDW

ATLSAT

FLL,MIA,PBIAUS

BWI,DCA,IADPHXDEN

SFO,SJC,OAKMCOLASHRLTPACRPPHLSANOKCTULBNAELPSTL

Seats (Millions)

Non

stop

Markets

IAH

HOU

024681012141618

Seats C

apacity (M

illions)

HOU (Houston/Hobby)IAH (Houston/Intercontinental)

294

Figure E-22: Non-Stop Markets Served in the Miami MAS – MIA* (Int’l), FLL* (Fort

Lauderdale), PBI (Palm Beach).

MIA (Year = 2008) FLL (Year = 2008)

PBI (Year = 2008)

295

Figure E-23: Non-Stop Markets Served in the Boston MAS

Figure E-24: Traffic Distribution by Market/Carrier among Airports in the Boston MAS.

BOS (Year = 2008)

MHT (Year = 2008)PVD (Year = 2008)

0.0 2.0 4.0

BWI,DCA,IADLGA,EWR,JFKORD,MDW

PHLFLL,MIA,PBI

ATLCLT

MCODTWSFO

LAX,LGBDFWTPACVGCLEMSPDENPHXLASIAHRDUPIT

Seats (Millions)

Non

stop

Markets

BOS

PVD

MHT

0.00.51.01.52.02.53.03.54.04.5

Seats C

apacity (M

illions)

MHT (Boston/Manchester)PVD (Boston/Providence)BOS (Boston/Logan)

296

E.3 Traffic Distribution by Market & Carrier among Airports in Boston Multi-Airport System

Figure E-25: Traffic Distribution by Market for FSCs 1 among Airports in MAS of Boston.

‐

1,000

2,000

3,000

4,000

5,000

6,000

Dep

artures Sche

duled

Non‐stop Markets Served by US Airways

BOS (Logan)

PVD (Providence)

MHT (Manchester)

0

1,000

2,000

3,000

4,000

5,000

6,000

Dep

artures S

ched

uled

Non‐stop Markets Served by Delta

BOS (Logan)

PVD (Providence)

MHT (Manchester)

0500

1,0001,5002,0002,5003,0003,5004,000

ORD DFW MIA LAX STL SFO LGA RDU JFK DCA CMHDep

artures S

ched

uled

Non‐stop Markets Served by American Airlines

BOS (Logan)PVD (Providence)MHT (Manchester)

297

Figure E-26: Traffic Distribution by Market for FSCs 2 among Airports in MAS of Boston.

0

500

1,000

1,500

2,000

2,500

3,000

3,500

ORD SFO IAD DEN LAXDep

artures S

ched

uled

Non‐stop Markets Served by United

BOS (Logan)

PVD (Providence)

MHT (Manchester)

0

1,000

2,000

3,000

4,000

5,000

EWR IAH CLEDep

artures S

ched

uled

Non‐stop Markets Served by Continental

BOS (Logan)

PVD (Providence)

MHT (Manchester)

0

500

1,000

1,500

2,000

2,500

3,000

DTW MSP MEM INDDep

artures S

ched

uled

Non‐stop Markets Served by Northwest

BOS (Logan)

PVD (Providence)

MHT (Manchester)

298

Figure E-27: Traffic Distribution by Market for LCCs among Airports in MAS of Boston.

0

1,000

2,000

3,000

4,000

5,000

BWI MCO PHL MDW TPA FLL LAS PHX BNADep

artures S

ched

uled


BOS (Logan)

PVD (Providence)

MHT (Manchester)

0

500

1,000

1,500

2,000

JFK MCO IAD FLL RSW PBI LGB TPA BUF RIC PITDep

artures S

ched

uled

Non‐stop Markets Served by JetBlue

BOS (Logan)

PVD (Providence)

MHT (Manchester)

0

500

1,000

1,500

2,000

2,500

BWI ATL PHF MCODep

artures S

ched

uled

Non‐stop Markets Served by AirTran

BOS (Logan)

PVD (Providence)

MHT (Manchester)

299

E.4 Average Flight Distance at Airports in Multi-Airport System

Figure E-28: Average Flight Distance at 44 Airports in 15 Multi-Airport Systems in the NAS

(Source: 2008 OAG).

0 500 1,000 1,500



HPN (New York/Westchester)SWF (New York/Stewart)

ORD (Chicago/O'Hare)MDW (Chicago/Midway)LAX (Los Angeles/Int'l)

















Average Flight Distance (Miles)

44 Airpo

rts in 15

Multi‐Airpo

rt Systems in the NAS

Red bar ‐ Primary airportBlue bar ‐ Secondary airport

300

Figure E-29: Southwest Airline Average Flight Distance during 1990-2008 (Source: 1990-2008

T-100 Segment).

Figure E-30: Southwest Airline Short-, Medium-haul Flight Share during 1990-2008 (Source:

1990-2008 T-100 Segment).

‐

100

200

300

400

500

600

700

Average

Flight Distance

Southwest Airlines

0%10%20%30%40%50%60%70%80%90%100%

Short‐ha

ul (<= 600 miles),

med

ium‐hau

l (600‐1200

miles) Flights S

hare

Long‐haul

Medium‐haul

Short‐haul

301

E.5 Connectivity at Secondary Airports in Multi-Airport System

Figure E-31: Top 5 OD Airport-Pairs of Connecting Passengers at MDW (Chicago/Midway)

Figure E-32: Top 5 OD Airport-Pairs of Connecting Passengers at DAL (Dallas/Love-Field)

Figure E-33: Top 5 OD Airport-Pairs of Connecting Passengers at HOU (Houston/Hobby).

302

E.6 Summary of Matlab Functions

Main_Summarize_T100.m Input T100I_SEGMENT_ALL_CARRIER_1990_2008_Enpl_by_Airport.txt (Predefined

input) T100D_SEGMENT_ALL_CARRIER_1990_2008_Enpl_by_Airport .txt (Predefined input) T100D_SEGMENT_ALL_CARRIER_1990_2008_by_Airport_Pair_Carrier.txt (Predefined input) CA_Enpls_Threshold (Predefined variable = 2,500) Airports_OEP.mat (Predefined input) OAG World Airports Database_Revised.txt (Predefined input) Proximity (Predefined variable = 60) Enplanements_Threshold (Predefined variable = 450,000) Share_Threshold_Secondary (Predefined variable = 0.01)

Output T100_Enpls_by_Airport.matT100_Enpls_by_Airport_CA.mat MAS.mat MAS_Short.mat MAS_Short_Transfered.mat T100D_Share_of_LCCs_by_Airport.mat

Summarize_T100_Enpls_by_Airport.mInput T100I_SEGMENT_ALL_CARRIER_1990_2008_Enpl_by_Airport.txt (Predefined

input) T100D_SEGMENT_ALL_CARRIER_1990_2008_Enpl_by_Airport .txt (Predefined input) CA_Enpls_Threshold (Predefined variable = 2,500)

Output T100_Enpls_by_Airport.matT100_Enpls_by_Airport_CA.mat

Identify_MAS.m Input Airports_OEP.mat (Predefined input)

T100_Enpls_by_Airport_CA.mat OAG World Airports Database_Revised.txt (Predefined input) Proximity (Predefined variable = 60) Enplanements_Threshold (Predefined variable = 450,000) Share_Threshold_Secondary (Predefined variable = 0.01)

Output MAS.mat MAS_Short.mat MAS_Short_Transfered.mat

303

Summarize_T100D_Share_of_LCCs_by_Airport.mInput Years_Analyzed (Predefined variable)

Airports_Analyzed (Predefined variable) T100D_SEGMENT_ALL_CARRIER_1990_2008_by_Airport_Pair_Carrier.txt (Predefined input) LCCs_US.mat (Predefined input)

Output T100D_Share_of_LCCs_by_Airport.matMain_Summarize_OAG.m Input Airports_Analyzed (Predefined variable)

OAG_Domestic_by_AirportPair_Carrier_2000/2004/2006/2008.txt (Predefined input) Airline_Hubs.mat (Predefined input) LCCs_US.mat (Predefined input) Flight_Range (Predefined variable = [600,1200]) OAG World Airports Database_Revised.txt (Predefined input) Departure_Threshold (Predefined variable = 365) MASs_US.mat

Output OAG_Summary_by_Carrier.matOAG_Carriers_by_Airport.mat OAG_Carriers_by_MAS.mat OAG_Markets_by_Airport.mat OAG_Markets_by_MAS.mat OAG_Markets_Combined_by_MAS.mat OAG_Markets_by_Carrier_by_MAS.mat OAG_Flight_Range_by_Airport.mat OAG_AC_Type_by_Airport.mat

Summarize_OAG_by_Carrier.m Input Case (Case == 1 % Summarize by seats capacity; Case == 2 Summarize

by aircraft operation) OAG_Domestic_by_AirportPair_Carrier_2000/2004/2006/2008.txt (Predefined input)

Output OAG_Summary_by_Carrier.mat

Summarize_OAG_Carriers_by_Airport.mInput Airports_Analyzed (Predefined variable)

OAG_Domestic_by_AirportPair_Carrier_2000/2004/2006/2008.txt (Predefined input) Departure_Threshold (Predefined variable = 365)

Output OAG_Carriers_by_Airport.mat

Summarize_OAG_Carriers_by_MAS.mInput MASs_US.mat

OAG_Carriers_by_Airport.matOutput OAG_Carriers_by_MAS.mat

Summarize_OAG_Markets_by_Carrier_by_MAS.mInput MASs_US.mat

OAG_Carriers_by_Airport.mat OAG_Carriers_by_MAS.mat

Output OAG_Markets_by_Carrier_by_MAS.mat

Summarize_OAG_Markets_by_Airport.mInput Airports_Analyzed (Predefined variable)

OAG_Domestic_by_AirportPair_Carrier_2000/2004/2006/2008.txt (Predefined input) Departure_Threshold (Predefined variable = 365)

304

Output OAG_Markets_by_Airport.matSummarize_OAG_Markets_by_MAS.mInput MASs_US.mat

OAG_Markets_by_Airport.matOutput OAG_Markets_by_MAS.mat

Summarize_OAG_Markets_Combined_by_MAS.mInput MASs_US.mat

OAG_Markets_by_MAS.matOutput OAG_Markets_Combined_by_MAS.mat

Summarize_OAG_Flight_Range_by_Airport.mInput Airports_Analyzed (Predefined variable)

OAG_Domestic_by_AirportPair_Carrier_2000/2004/2006/2008.txt (Predefined input) Airline_Hubs.mat (Predefined input) LCCs_US.mat (Predefined input) Flight_Range (Predefined variable = [600,1200]) OAG World Airports Database_Revised.txt (Predefined input) Departure_Threshold (Predefined variable = 365)

Output OAG_Flight_Range_by_Airport.mat

Summarize_OAG_AC_Type_by_Airport.mInput Airports_Analyzed (Predefined variable)

OAG_Domestic_by_AirportPair_Carrier_2000/2004/2006/2008.txt (Predefined input) Airline_Hubs.mat (Predefined input) LCCs_US.mat (Predefined input) Departure_Threshold (Predefined variable = 365)

Output OAG_AC_Type_by_Airport.mat

Main_DB1B_OD_Connectivity.m Input MASs_US.mat

Airports_OEP_not_in_MAS.mat (Predefined input) OAG World Airports Database_Revised.txt (Predefined input)

Output DB1B_OD_Passengers_at_Airport_in_MAS.matDB1B_OD_Passengers_at_Airport_not_in_MAS.mat DB1B_OD_Markets_by_MAS.mat DB1B_OD_Passengers_Share_by_Airport.mat DB1B_Connectivity_at_Airport_in_MAS.mat

The following functions are the sub-functions for Main_DB1B_OD_Connectivity.m Summarize_DB1B_OD_Passengers_by_Airport_by_Q.mInput MASs_US.mat

DB1B_D_MARKET_Y2008_Q1.txt (Predefined input) DB1B_D_MARKET_Y2008_Q2.txt (Predefined input) DB1B_D_MARKET_Y2008_Q3.txt (Predefined input) DB1B_D_MARKET_Y2008_Q4.txt (Predefined input)

Output OD_Passengers_MAS.mat

Summarize_DB1B_Connectivity_by_Airport_by_Q.mInput MASs_US.mat

DB1B_D_MARKET_Y2008_Q1.txt (Predefined input) DB1B_D_MARKET_Y2008_Q2.txt (Predefined input) DB1B_D_MARKET_Y2008_Q3.txt (Predefined input) DB1B_D_MARKET_Y2008_Q4.txt (Predefined input)

Output Connectivity.mat

305

Summarize_DB1B_OD_Passengers_by_Airport.mInput MASs_US.mat

OD_Passengers_MAS.mat Output DB1B_OD_Passengers_at_Airport_in_MAS.mat

Summarize_DB1B_OD_Passengers_by_Airport.mInput Airports_OEP_not_in_MAS.mat

OD_Passengers_MAS.mat Output DB1B_OD_Passengers_at_Airport_not_in_MAS.mat

Summarize_DB1B_OD_Markets_by_MAS.mInput MASs_US.mat

DB1B_OD_Passengers_at_Airport_in_MAS.matOutput DB1B_OD_Markets_by_MAS.mat

DB1B_OD_Passengers_Share_by_Airport.mat

Summarize_DB1B_Connectivity_by_Airport.mInput MASs_US.mat

Connectivity.mat OAG World Airports Database_Revised.txt (Predefined input)

Output Connectivity.mat

Draw_OAG_Markets_by_Airport.mInput OAG World Airports Database_Revised.txt (Predefined input)

Departure_Threshold (Predefined variable = 365) OAG_Domestic_by_AirportPair_Carrier_2000/2004/2006/2008.txt Airline_Hubs.mat (Predefined input) LCCs_US.mat (Predefined input)

Sample Output

MDW (Year = 2008)

306

Draw_Connecting_Airports_by_Airport.mInput OAG World Airports Database_Revised.txt (Predefined input)

Departure_Threshold (Predefined variable = 365) OAG_Domestic_by_AirportPair_Carrier_2000/2004/2006/2008.txt (Predefined input) Airline_Hubs.mat (Predefined input) LCCs_US.mat (Predefined input) DB1B_Connecting_OD_Pairs_at_Airport_in_MAS.mat Connecting_Passengers_Coverage (Predefined variable)

Sample Output

Draw_Connecting_OD_Pairs_by_Airport.mInput OAG World Airports Database_Revised.txt (Predefined input)

Departure_Threshold (Predefined variable = 365) OAG_Domestic_by_AirportPair_Carrier_2000/2004/2006/2008.txt (Predefined input) Airline_Hubs.mat (Predefined input) LCCs_US.mat (Predefined input) DB1B_Connecting_OD_Pairs_at_Airport_in_MAS.mat Connecting_Passengers_Coverage (Predefined variable)

Sample Output

307

E.7 Description of Input and Output Variables for Matlab Functions

Input/Output Variable Dimension of

Variable Remark on Dimension of Variable

T100_Enpls_by_Airport.mat 611 (No. of airports) x 22

1: Airport 2‐21: 1990‐2008 enplanements 22: Min. of 1990‐2008 enplanements

T100_Enpls_by_Airport_CA.mat 486 (No. of commercial airports) x 22

1: Commercial airport with at least 2,500 enplanements 2‐21: 1990‐2008 enplanements 22: Min. of 1990‐2008 enplanements

MAS.mat

28 (No. of metropolitan

areas including 34

OEP airports) x 6

1: OEP airports in the metropolitan area 2: Total enplanements 3: No. of airports x 20

1: Airport 2‐19: 1990‐2008 enplanements

4: No. of airports x 20 1: Airport 2‐19: 1990‐2008 enplanements share

3: No. of active airports x 20 1: Active airport 2‐19: 1990‐2008 enplanements

3: No. of active airports x 20 1: Active airport 2‐19: 1990‐2008 enplanements share

MAS_Short.mat 15 (No. of MASs) x 6

1: OEP airports in the MAS 2: Total enplanements 3: No. of airports x 20

1: Airport 2‐19: 1990‐2008 enplanements

4: No. of airports x 20 1: Airport

308

2‐19: 1990‐2008 enplanements share 3: No. of active airport x 20

1: Active airport 2‐19: 1990‐2008 enplanements

3: No. of active airports x 20 1: Active airport 2‐19: 1990‐2008 enplanements share

MAS_Short_Transferred.mat

2 (1‐Enplanements;

2‐Enplanements share in MAS) x 44 (No. of airports in MAS) x 21

1: MAS Indices 2: Airports in MAS 3‐21: 1990‐2008 enplanements/enplanements share in MAS

T100D_Share_of_LCCs_by_Airport.mat 44 (No. of airports in MAS) x 3

1: Airport 2: No. of LCCs x 21

1: LCC ID 2: LCC Name 3‐21: 1990‐2008 enplanements

3: (No. of LCCs + 1) x 21 1: LCC ID 2: LCC Name 3‐21: 1990‐2008 enplanements share

OAG_Summary_by_Carrier.mat 2 (1‐Seats; 2‐Departures) x No. of Carriers

1: Carrier ID 2: Carrier name 3: Total seats/departures 4: No. of airports serving x 8

1: Airport 2: No. of markets served at the airport 3: Seats/Departures

309

4: No. of markets served at the airport x 2 1: Market 2: Seats/Departures

OAG_Carriers_by_Airport.mat 44 (No. of airports in MAS) x 4

1: Airport 2: Seats 3: Departures 4: No. of carriers serving the airport x 6

1: Carrier ID 2: Carrier Name 3: Seats 4: Departures 5: Seats share over the total seats at the airport 6: No. of markets served by the carrier x 5

1: Market 2: Seats 3: Departures 4: Seats share of total seats provided by the carrier 5: No. of aircraft type x 3

1: Seats capacity of the aircraft type 2: Category of aircraft type 3: Departures

OAG_Carriers_by_MAS.mat 15 (No. of MASs) x 4

*N = No. of airports in the MAS 1: MAS Index 2: Active airports in the MAS 3: No. of carriers in the MAS x (2*N + 4)

1: Carrier ID 2: Carrier Name 3‐(N +2): Seats at the airport N+3: Total seats in the MAS (N+4)‐(2*N +3): Departures at the airport

310

2*N + 4: Total departures in the MAS 4: No. of carriers in the MAS x (2*N + 2)

1: Carrier ID 2: Carrier Name 3‐(N +2): Seats share at the airport (N+3)‐(2*N +2): Departures share at the airport

OAG_Markets_by_Airport.mat 44 (No. of airports in MAS) x 4

1: Airport 2: Seats 3: Departures 4: No. of markets served at the airport x 5

1: Market 2: Seats 3: Departures 4: Seats share over the total seats at the airport 5: No. of carriers in the market x 5

1: Carrier ID 2: Carrier Name 3: Seats 4: Departures 5: Seats share of total seats for the market

OAG_Markets_by_MAS.mat 15 (No. of MASs) x 4

*N = No. of airports in the MAS 1: MAS Index 2: Active airports in the MAS 3: No. of combined markets in the MAS x (2*N + 3)

1: Market 2‐(N +1): Seats at the airport N+2: Total seats in the MAS (N+3)‐(2*N +2): Departures at the airport 2*N + 3: Total departures in the MAS

4: No. of combined markets in the MAS x (2*N + 1) 1: Market

311

2‐(N +1): Seats share at the airport (N+2)‐(2*N +1): Departures share at the airport

OAG_Markets_Combined_by_MAS.mat 15 (No. of MASs) x 4

*N = No. of airports in the MAS 1: MAS Index 2: Active airports in the MAS 3: No. of combined markets in the MAS x (2*N + 4)

1: Combined market 2‐(N +1): Seats at the airport N+2: Total seats in the MAS (N+3)‐(2*N +2): Departures at the airport 2*N + 3: Total departures in the MAS 2*N + 4: No. of markets combined x (2*N +3)

1: Market 2‐(N +1): Seats at the airport N+2: Total seats in the MAS (N+3)‐(2*N +2): Departures at the airport 2*N + 3: Total departures in the MAS

OAG_Markets_by_Carrier_by_MAS.mat 15 (No. of MASs) x 4

*N = No. of airports in the MAS 1: MAS Index 2: Active airports in the MAS 3: No. of carrier‐market combination x (2*N + 4)

1: Carrier ID 2: Market 3‐(N +2): Seats at the airport N +3: Total seats at the airport (N+4)‐(2*N +3): Departures at the airport 2*N + 4: Total departures in the MAS

OAG_Flight_Range_by_Airport.mat

3 (1‐All carriers; 2‐FSCs; 3‐LCCs) x 44 (No. of

1: Airport 2: Total departures 3: Average flight range 4: Share of low‐haul (<600 miles)

312

airports in MAS) x 6

5: Share of medium‐haul (600‐1200 miles) 6: Share of long‐haul (>1200 miles)

OAG_AC_Type_by_Airport.mat

3 (1‐All carriers; 2‐FSCs; 3‐LCCs) x 44 (No. of airports in MAS) x 9

1: Airport 2: Total departures 3: Average seats capacity per departure 4: Share of Pistol (7‐9 seats) 5: Share of Helicopter (8 seats) 6: Share of Turboprop (19‐74 seats) 7: Share of Narrow‐body Jets (37‐99 seats) 8: Share of Narrow‐body Jets (100‐252 seats) 9: Share of Wide‐body Jets (165‐397 seats)

DB1B_OD_Passengers_at_Airport_in_MAS.mat 44 (No. of airports in MAS) x 5

1: MAS Index 2: Airport 3: Primary/secondary airport ID 4: O&D passengers 5: No. of O&D markets x 2

1: O&D market 2: O&D passengers

DB1B_OD_Passengers_at_Airport_not_in_MAS.mat 13 (No. of

airports not in MAS) x 5

1: MAS Index 2: Airport 3: Primary/secondary airport ID 4: O&D passengers 5: No. of O&D markets x 2

1: O&D market 2: O&D passengers

DB1B_OD_Passengers_Share_by_Airport.mat 44 (No. of airports in MAS) x 2

1: Airport 2: O&D passenger share in the MAS

Modeling of Airline and Passenger Dynamics in the National ... · Modeling of Airline and Passenger...

Documents

Transcript of Modeling of Airline and Passenger Dynamics in the National ... · Modeling of Airline and Passenger...