Flght Delays, Capacty Investment and Welfare under Ar Transport Supply-demand Equlbrum Bo Zou 1, Mark Hansen 2 1 Unversty of Illnos at Chcago 2 Unversty of Calforna at Berkeley
2
Total economc mpact of flght delay: 3
Total economc mpact of flght delay: $32 bllon n 2007 4
Total economc mpact of flght delay: $32 bllon n 2007 Ball, M., Barnhart, C., Dresner, M., Hansen, M. Neels, K., Odon, A., Peterson, E., Sherry, L., Tran, A., Zou, B., 2010. Total Delay Impact Study: A Comprehensve Assessment of the Costs and Impacts of Flght Delay n the Unted States. Major Partcpant. Report Prepared for the US Federal Avaton Admnstraton. Zou, B., Hansen, M., 2012. Impact of Operatonal Performance on Ar Carrer Cost Structure: Evdence from US Arlnes. Transportaton Research Part E: Logstcs and Transportaton Revew, 48 (6), 1032-1048. Hansen, M., Zou, B., 2013. Arport Operatonal Performance and ts Impact on Arlne Cost. In: Odon, A. and Zografos, K. (eds.), Modelng and Managng Arport Performance: Theory and Practce, John Wley & Sons, Inc. 5
Means to mtgate flght delay Managng demand 6
Means to mtgate flght delay Managng demand Congeston prcng 7
Means to mtgate flght delay Managng demand Congeston prcng Slot control 8
Means to mtgate flght delay Managng demand Congeston prcng Slot control Swaroop, P., Zou, B., Ball, M., Hansen, M., 2012. Do More U.S. Arports Need Slot Controls? A Welfare Based Approach to Determne Slot Levels. Transportaton Research Part B: Methodologcal, (9), 1239-1259. Ball, M., Hansen, M., Swaroop, P., Zou, B., 2013. Desgn and Justfcaton for Market-Based Approaches to Arport Congeston Management. In: Odon, A. and Zografos, K. (eds.), Modelng and Managng Arport Performance: Theory and Practce, John Wley & Sons, Inc. 9
Means to mtgate flght delay Managng demand Congeston prcng Slot control Swaroop, P., Zou, B., Ball, M., Hansen, M., 2012. Do More U.S. Arports Need Slot Controls? A Welfare Based Approach to Determne Slot Levels. Transportaton Research Part B: Methodologcal, (9), 1239-1259. Ball, M., Hansen, M., Swaroop, P., Zou, B., 2013. Desgn and Justfcaton for Market-Based Approaches to Arport Congeston Management. In: Odon, A. and Zografos, K. (eds.), Modelng and Managng Arport Performance: Theory and Practce, John Wley & Sons, Inc. Increasng supply 10
Outlne Background Research Framework Equlbrum Model Concluson 11
Background Framework Framework Model 1 Model Model 2 Concluson Investment Infrastructure Capacty ( ) Benefts Arlne Cost Passenger travel tme ( ) ( ) Flght Delay ( ) 12
Background Framework Framework Model 1 Model Model 2 Concluson Issues wth the approach Ceters parbus assumpton Predctng future Investment Infrastructure Capacty ( ) Benefts Arlne Cost Passenger travel tme ( ) ( ) Flght Delay ( ) 13
Background Framework Framework Model 1 Model Model 2 Concluson Objectve of the Research Develop an nnovatve methodology to systematcally capture supply-demand response to nvestment 14
Background Framework Framework Model 1 Model Model 2 Concluson Outlne Background Research Framework Equlbrum Models Arlne competton model User equlbrum model Concluson 15
Background Framework Framework Model 1 Model Model 2 Concluson No congeston Hgher densty Hgher flght frequency Lower unt cost Improved servce qualty Lower fare Reduced generalzed cost More demand 16
Background Framework Framework Model 1 Model Model 2 Concluson Wth congeston Hgher densty Hgher flght frequency Lower unt cost Improved servce qualty Lower fare Reduced generalzed cost More demand 17
Background Framework Framework Model 1 Model Model 2 Concluson Demand Unconstraned user supply Demand (passenger-mles) 18
Background Framework Framework Model 1 Model Model 2 Concluson Demand Constraned user supply Unconstraned user supply Demand (passenger-mles) 19
Background Framework Framework Model 1 Model Model 2 Concluson Demand Constraned user supply Constraned user supply after nvestment Unconstraned user supply Demand (passenger-mles) 20
Background Framework Framework Model 1 Model Model 2 Concluson Demand Constraned user supply Constraned user supply after nvestment Unconstraned user supply Demand (passenger-mles) 21
Background Framework Framework Model 1 Model Model 2 Concluson Conventonal vew Investment Infrastructure Capacty Flght Delay 22
Background Framework Framework Model 1 Model Model 2 Concluson Proposed framework Investment Passenger Demand Flght Traffc Infrastructure Capacty Arfare Flght Delay Arlne Cost 23
Background Framework Framework Model 1 Model Model 2 Concluson Proposed framework Investment Passenger Demand Flght Traffc Infrastructure Capacty Arfare Flght Delay Maxmzng Proft Arlne Cost Output 24
Background Framework Framework Model 1 Model Model 2 Concluson Proposed framework Investment Passenger Demand Flght Traffc Infrastructure Capacty Arfare Flght Delay Maxmzng Proft Arlne Cost Output 25
Background Framework Framework Model 1 Model Model 2 Concluson Outlne Background Research Framework Equlbrum Models Concluson 26
Background Framework Framework Model 1 Model Model 2 Concluson Outlne Background Research Framework Equlbrum Models Arlne competton model User equlbrum model Concluson Zou, B., Hansen, M., 2012. Flght Delays, Capacty Investment and Socal Welfare under Ar Transport Supply-Demand Equlbrum. Transportaton Research Part A: Polcy and Practce 46 (6), 965-980.
Demand Supply Equlbrum Equlbrum shft Consder a duopoly market 28
Demand Supply Equlbrum Equlbrum shft Consder a duopoly market Utlty of a representatve ndvdual 00 1 1 2 U( q0, q1, q2 ) q0 ( q1 q2 ) ( 01q1 2 2 2 02q1q2 01q 2 01 02 01 02 2 2 ) 29
Demand Supply Equlbrum Equlbrum shft Consder a duopoly market Utlty of a representatve ndvdual 00 1 1 2 U( q0, q1, q2 ) q0 ( q1 q2 ) ( 01q1 2 2 2 02q1q2 01q 2 Consumpton of numerare goods 01 02 01 02 2 2 ) 30
Demand Supply Equlbrum Equlbrum shft Consder a duopoly market Utlty of a representatve ndvdual 00 1 1 2 U( q0, q1, q2 ) q0 ( q1 q2 ) ( 01q1 2 2 2 02q1q2 01q 2 Consumpton of arlne 1 s servce 01 02 01 02 2 2 ) 31
Demand Supply Equlbrum Equlbrum shft Consder a duopoly market Utlty of a representatve ndvdual 00 1 1 2 U( q0, q1, q2 ) q0 ( q1 q2 ) ( 01q1 2 2 2 02q1q2 01q 2 Consumpton of arlne 2 s servce 01 02 01 02 2 2 ) 32
Demand Supply Equlbrum Equlbrum shft Consder a duopoly market Utlty of a representatve ndvdual 00 1 1 2 U( q0, q1, q2 ) q0 ( q1 q2 ) ( 01q1 2 2 2 02q1q2 01q 2 01 02 01 02 2 2 ) 02: parameters ( 01 ) 00, 01, 02 33
Demand Supply Equlbrum Equlbrum shft max U( q0, q1, q2) s. t. q 0 Pq 1 1 P2 q2 I 34
Demand Supply Equlbrum Equlbrum shft max U( q0, q1, q2) s. t. q 0 Pq 1 1 P2 q2 I Generalzed cost for choosng arlne 1 35
Demand Supply Equlbrum Equlbrum shft max U( q0, q1, q2) s. t. q 0 Pq 1 1 P2 q2 I Generalzed cost for choosng arlne 2 36
Demand Supply Equlbrum Equlbrum shft max U( q0, q1, q2) s. t. P q 0 Pq 1 1 P2 q2 P kl f I 1,2 37
Demand Supply Equlbrum Equlbrum shft max U( q0, q1, q2) s. t. P q 0 Pq 1 1 P2 q2 P f kl I 1,2 Arfare 38
Demand Supply Equlbrum Equlbrum shft max U( q0, q1, q2) s. t. P q 0 Pq 1 1 P2 q2 P f kl I 1,2 Arfare Schedule delay 39
Demand Supply Equlbrum Equlbrum shft max U( q0, q1, q2) s. t. P q 0 Pq 1 1 P2 q2 P f kl I 1,2 Arfare Schedule delay Delay 40
Demand Supply Equlbrum Equlbrum shft max U( q0, q1, q2) Composte of ncome and travel tme constrants s. t. P q 0 Pq 1 1 P2 q2 P f kl I 1,2 Arfare Schedule delay Delay 41
Demand Supply Equlbrum Equlbrum shft Indvdual demand q 01 02 ( ) 00 01P 02P 01 02 kl, 1,2 f f ( ) 02 01 42
Demand Supply Equlbrum Equlbrum shft q Indvdual demand 01 02 ( ) 00 01P 02P 01 02 kl, 1,2 f f ( ) 02 Market demand 01 Q 1 2 0 1P 2P L, 1,2 f f ( 1 2) 43
Demand Supply Equlbrum Equlbrum shft Flght operatng cost for trp C c 0 s s L 44
Demand Supply Equlbrum Equlbrum shft Flght operatng cost for trp C c 0 s s L Fxed cost 45
Demand Supply Equlbrum Equlbrum shft Flght operatng cost for trp C c 0 s s L Fxed cost Varable cost as a functon of arcraft sze s 46
Demand Supply Equlbrum Equlbrum shft Flght operatng cost for trp C c 0 s s L Fxed cost Varable cost as a functon of arcraft sze s Delay cost as a functon of arcraft sze s and delay L 47
Demand Supply Equlbrum Equlbrum shft Flght operatng cost for trp C c 0 s s L Fxed cost Varable cost as a functon of arcraft sze s Delay cost as a functon of arcraft sze s and delay L 48
Demand Supply Equlbrum Equlbrum shft Assumpton: each flght s full 49
Demand Supply Equlbrum Equlbrum shft Assumpton: each flght s full Q f s Passenger demand Flght frequency Arcraft sze 50
Demand Supply Equlbrum Equlbrum shft max P Q f C for 1,2 51
Demand Supply Equlbrum Equlbrum shft max P Q f C for 1,2 Arfare Passenger demand Flght frequency Operatng cost per flght 52
Demand Supply Equlbrum Equlbrum shft Assume arlnes compete on fare and frequency smultaneously n a Nash fashon 0 0 1,2 P f 53
Demand Supply Equlbrum Equlbrum shft Assume arlnes compete on fare and frequency smultaneously n a Nash fashon Symmetrc arlnes 0 0 1,2 P f P P P 1 2 f f f 1 2 54
2 1 1 2 1 2 1 2 1 2 1 1 0 2 2 2 ) ( 2 L L f P 55 Demand Supply Equlbrum Prce response Equlbrum shft
2 1 1 2 1 2 1 2 1 2 1 1 0 2 2 2 ) ( 2 L L f P 56 Demand Supply Equlbrum Prce response Constant Equlbrum shft
Demand Supply Equlbrum Equlbrum shft Prce response P ( 1 2) 0 1 f L 1L 2 2 2 2 1 2 1 2 1 2 1 2 Constant Frequency effect on WTP 57
Demand Supply Equlbrum Equlbrum shft Prce response P ( 1 2) 0 1 f L 1L 2 2 2 2 1 2 1 2 1 2 1 2 Constant Frequency effect on WTP Delay effect on WTP 58
Demand Supply Equlbrum Equlbrum shft Prce response P ( 1 2) 0 1 f L 1L 2 2 2 2 1 2 1 2 1 2 1 2 Constant Frequency effect on WTP Delay effect on WTP Arlne delay cost passed onto passengers 59 59
Demand Supply Equlbrum Equlbrum shft Compare equlbrum wth and wthout congeston 60
Demand Supply Equlbrum Equlbrum shft Compare equlbrum wth and Wth congeston Frequency ( ) wthout congeston 61
Demand Supply Equlbrum Equlbrum shft Compare equlbrum wth and Wth congeston Frequency ( ) wthout congeston Passenger generalzed cost ( ) 62
Demand Supply Equlbrum Equlbrum shft Compare equlbrum wth and Wth congeston Frequency ( ) wthout congeston Passenger generalzed cost ( ) Passenger demand ( ) 63
Demand Supply Equlbrum Equlbrum shft Compare equlbrum wth and Wth congeston Frequency ( ) wthout congeston Passenger generalzed cost ( ) Passenger demand ( ) Fare (?) Arcraft sze (?) Unt operatng cost per passenger (?) 64
Demand Supply Equlbrum Equlbrum shft Smulaton analyss 65
Demand Supply Equlbrum Equlbrum shft Smulaton analyss Assumpton about arport delay L Delay on a market s determned by the more congested arport N ndependent and dentcal markets nto that arport 66
Demand Supply Equlbrum Equlbrum shft Smulaton analyss Assumpton about arport delay L L [ N( f f K )/ ], 1 2 1 Total arport traffc Arport capacty 67
Demand Supply Equlbrum Equlbrum shft Smulaton analyss Assumpton about arport delay L L [ N( f K 1 f2)/ ], 1 All other parameters derved from emprcal evdence 68
Demand Supply Equlbrum Equlbrum shft Scenaros Fare Arcraft sze Unt operatng cost ($/passenger) Infnte capacty (no delay) Fnte capacty (720 operatons per day, wth delay) 98.9 63.6 91.4 96.0 71.9 91.5 69
Demand Supply Equlbrum Equlbrum shft Scenaros Fare Arcraft sze Unt operatng cost ($/passenger) Infnte capacty (no delay) Fnte capacty (720 operatons per day, wth delay) 98.9 63.6 91.4 96.0 71.9 91.5 Decreased WTP domnates arlnes tendency to pass part of the delay cost to passengers 70
Demand Supply Equlbrum Equlbrum shft Scenaros Fare Arcraft sze Unt operatng cost ($/passenger) Infnte capacty (no delay) Fnte capacty (720 operatons per day, wth delay) 98.9 63.6 91.4 96.0 71.9 91.5 Use larger planes to avod hgh delays 71
Demand Supply Equlbrum Equlbrum shft Scenaros Fare Arcraft sze Unt operatng cost ($/passenger) Infnte capacty (no delay) Fnte capacty (720 operatons per day, wth delay) 98.9 63.6 91.4 96.0 71.9 91.5 Delay cost partally offset by economes of arcraft sze 72
Demand Supply Equlbrum Equlbrum shft Comparson between equlbrum and conventonal approaches 73
Demand Supply Equlbrum Equlbrum shft Increase arport capacty by 50% 74
Demand Supply Equlbrum Equlbrum shft Increase arport capacty by 50% Conventonal 5.6 Equlbrum 4.7 Arport delay savng (mn/flght) 75
Demand Supply Equlbrum Equlbrum shft Increase arport capacty by 50% Conventonal 5.6 Equlbrum 4.7 Equlbrum 163 Conventonal 70 Arport delay savng (mn/flght) Consumer surplus (mllon$) 76
Outlne Background Research Framework Equlbrum Models Arlne competton model User equlbrum model Concluson 77
Introducton of basc concepts Route Segment Market Non-stop Route Segment AB Spoke cty B Segment CB One-stop Route Spoke cty A Segment AC Hub cty C 78
Demand Supply Equlbrum Equlbrum shft Demand estmaton 79
Demand NL3B Supply Equlbrum Equlbrum shft Scale λ m Nest structure Ar Non-Ar Scale λ a O-D Arport Scale λ p O 1 O 1 H 1 D 1 O 1 H 2 D 1 O 1 H h D 1 D 1 O 1 D 2 O 1 H 1 D 2 O 1 H 2 D 2 O 1 H h D 2 O o D n O o H 1 D n O o H 2 D n O o H h D n Source: Hsao, C. and Hansen, M., 2011. A passenger demand model for ar transportaton n a hub-and-spoke network. Transportaton Research Part E: Logstcs 80 and Transportaton Revew, Vol. 47, 1112-1125.
Demand Supply Equlbrum Equlbrum shft User equlbrum formulaton Demand = G 1 (Fare, Flght Traffc, Arport delay) s.t. Constrants 81
Demand Supply Equlbrum Equlbrum shft Investment Passenger Demand Flght Traffc Infrastructure Capacty Arfare Flght Delay Maxmzng Proft Arlne Cost Output 82
Demand Supply Equlbrum Equlbrum shft Flght traffc = G 2 (Demand, Arport Delay) Fare = G 3 (Demand, Arport delay) Arport delay= G 4 (Flght traffc) Zou, B., Hansen, M. Flght Delay Impact on Arfare and Flght Frequency: A Comprehensve Assessment. Paper to be submtted to Transportaton Research Part 83 A.
Demand Supply Equlbrum Equlbrum shft (++) (-) Flght traffc = G 2 (Demand, Arport delay) (+/-) (+) Fare = G 3 (Demand, Arport delay) (++) Arport delay= G 4 (Flght traffc) 84
Demand Supply Equlbrum Equlbrum shft User equlbrum formulaton Demand = G 1 (Fare, Flght Traffc, Arport delay) s.t. Flght traffc = G 2 (Demand, Arport delay) Fare = G 3 (Demand, Arport delay) Arport delay= G 4 (Flght traffc) 85
Demand Supply Equlbrum Equlbrum shft Smultaneous equaton system Demand = G 1 (Fare, Flght Traffc, Arport delay) Flght traffc = G 2 (Demand, Arport delay) Fare = G 3 (Demand, Arport delay) Arport delay= G 4 (Flght traffc) 86
Demand Supply Equlbrum Equlbrum shft Smulaton analyss 87
Demand Supply Equlbrum Equlbrum shft Network S 50 S 1 S 2 S 3 H 400 mles S S j 88
Demand Supply Equlbrum Equlbrum shft Network S 50 S 1 S 2 S 3 H 400 mles S S j 89
Demand Demand Supply Equlbrum Equlbrum shft 8000 6000 0-stop 1-stop Spoke-spoke market demand 4000 2000 0 0 200 400 600 800 90 O-D Dstance (mles)
Demand Demand Supply Equlbrum Equlbrum shft 8000 0-stop 1-stop Spoke-spoke market demand 6000 4000 Total demand s low because the dstance s too short 2000 0 0 200 400 600 800 91 O-D Dstance (mles)
Demand Demand Supply Equlbrum Equlbrum shft 8000 6000 4000 2000 0-stop 1-stop Spoke-spoke market demand 1-stop routes more attractve because of reduced crcuty (actually dstance/ OD dstance) 0 0 200 400 600 800 92 O-D Dstance (mles)
Flghts/quarter Demand Supply Equlbrum Equlbrum shft 250 Spoke-spoke segment frequency 200 150 100 50 0 0 200 400 600 800 Segment Dstance (mles) 93
Demand Supply Equlbrum Equlbrum shft Delay (mn/flght) Hub Spoke Before 26.5 11.6 94
Demand Supply Equlbrum Equlbrum shft Increase hub capacty by 50% Delay (mn/flght) Hub Spoke Before 26.5 11.6 95
Demand Supply Equlbrum Equlbrum shft Increase hub capacty by 50% Delay (mn/flght) Hub Spoke Before 26.5 11.6 After 17.5 11.4 96
Demand Demand Supply Equlbrum Equlbrum shft 8000 6000 Spoke-spoke market demand shft 0-stop (before) 1-stop (before) 0-stop (after) 1-stop (after) 4000 2000 0 0 200 400 600 97800 O-D Dstance (mles)
Total Demand Demand Supply Equlbrum Equlbrum shft 12000 10000 Before After Total spoke-spoke market demand 8000 6000 4000 2000 0 0 200 400 600 800 98 O-D Dstance
Total Demand Demand Supply Equlbrum Equlbrum shft 12000 10000 8000 Before After Total spoke-spoke market demand Induced demand 6000 4000 2000 0 0 200 400 600 800 99 O-D Dstance
Flghts/quarter Demand Supply Equlbrum Equlbrum shft Spoke-spoke segment frequency change 250 Before 200 After 150 100 50 0 0 200 400 600 800 Segment Dstance (mles)
Flghts/quarter Demand Supply Equlbrum Equlbrum shft Spoke-hub segment frequency change 2500 2000 1500 1000 500 0 Before After
$ Demand Supply Equlbrum Equlbrum shft Consumer surplus change per ar travel decson makng 0.004 0.0035 0.003 0.0025 0.002 0.0015 0.001 0.0005 0-0.0005 Negatve 0 200 400 600 800 Dstance (mles) 102
Demand Supply Equlbrum Equlbrum shft Comparson between equlbrum and conventonal approaches 103
Background Framework Model Concluson Other Research Conventonal 14.2 Demand Supply Equlbrum Equlbrum shft Equlbrum 9.0 Hub delay savngs (mn/flght) 104
Background Framework Model Concluson Other Research Demand Supply Equlbrum Equlbrum shft Conventonal 14.2 Equlbrum 218.4 Equlbrum 9.0 Conventonal 69.5 Hub delay savngs (mn/flght) Passenger welfare gan (mllon$/qtr) 105
Outlne Background Research Framework Equlbrum Models Arlne competton model User equlbrum model Concluson 106
Summary An equlbrum framework 107
Summary An equlbrum framework Larger and broader benefts 108
Summary An equlbrum framework Larger and broader benefts Addtonal nsghts Delay trggers nvestment 109
Summary An equlbrum framework Larger and broader benefts Addtonal nsghts Delay trggers nvestment Returns more than delay savngs 110
Summary An equlbrum framework Larger and broader benefts Addtonal nsghts Delay trggers nvestment Returns more than delay savngs Delay reducton less than expected 111
Summary An equlbrum framework Larger and broader benefts Addtonal nsghts Delay trggers nvestment Returns more than delay savngs Delay reducton less than expected Investment paradox: some markets can be worse off 112
Extensons 113
Extensons Infrastructure nvestment decson makng Sze, locaton, tmng 114
Extensons Infrastructure nvestment decson makng Sze, locaton, tmng Envronmental externaltes 115
Extensons Infrastructure nvestment decson makng Sze, locaton, tmng Envronmental externaltes Consder ntermodal competton 116
Thank you! Questons? 117