Optimal Air Traffic Delays Miguel Urdanoz SESAR 1 st December 2011 - PowerPoint PPT Presentation

Optimal Air Traffic Delays Miguel Urdanoz SESAR 1 st December 2011 Joint work with Marc Ivaldi, Emile Quinet and Etienne de Villemeur

���������������������������������������� �������������������������� Source: L’observatoire des retards du transport aérien � 2 2 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011 2

����������� � European Commission (2005) " Strengthening passenger rights within the European Union “ � Increase monetary compensations for denied boarding � Includes compensations for some kind of delays Includes compensations for some kind of delays � Include compensations for long delays � Airlines: increase in costs that will be translated to an increase in price. 3 3 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011 3

��� ������ � How do we measure delays? � Are delays so bad? Are delays so bad? 4 4 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

�������������������� � Difference between scheduled arrival time and real arrival time � Buffer time or buffer delays: extra time added to the minimum required travel time � Few studies about buffer time: � Morrison, Winston, Bailey and Khan (1989) � Mayer and Sinai (2003) 5 5 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

����������� ( ) φ ε ξ ε T ij Buffer time Observed delays 6 6 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

����������������� Westminster ITA Madrid Airport Study Madrid Airport July Market and time coverage Europe 1999 Europe 2004 1997-2000 Airlines Passengers and Passengers and Studied costs airlines airlines Schedule and Schedule and Delays Schedule buffer buffer 2364-2916 5000 � /hour 4320 � /hour Airlines � /hour Estimated Costs 44.6-60 � /hour 15,9 � /hour Passengers 7 7 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

������������ ������������������� � Modeling the queues due to congestion � Brueckner (2002): carriers internalize the congestion they impose in themselves � Mayer and Sinai (2003): Delays appear due to network benefits from hubbing and congestion network benefits from hubbing and congestion externalities. � To study delay costs, we should not consider the whole delay. 8 8 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

����!��������������������� � Optimal delay � Value of delays that maximizes social welfare � Social welfare = Firm’s profits + Consumer Surplus � Equilibrium delay � Value of delays that maximizes firm’s profit 9 9 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

���������� � Step 1: Computation of equilibrium delays � The observed data are assumed to obey the equilibrium conditions � Invert the relationship to obtain the parameters of demand � Recover welfare function from the demand function � Step 2: Computation of optimal delays � Maximization of welfare � Step 3: Evaluation of the cost of delays � Welfare at optimum@Welfare at equilibrium 10 10 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

������������������������������������������������� ���������������" 11 11 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

��������������������������������������� � Hub@and@spokes network ( ) ( ) = + α + β C X F X T ij ij ij ij � Stochastic delay ( ) ∼ ε Φ ε ij ij � Airline introduce to control for delays Airline introduce to control for delays ζ ij � Passengers connect at the hub � Airline introduces δ and can introduce γ γ ≤ δ 12 12 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

���������������� � Firm is a monopoly � Faces 6 demands   ( ) ∞ ∫ ( ) ( )   = + + + ζ + ε − ζ φ ε ε X a b P v T r d     12 12 12 12 12 12 12 12 ζ + sd 12 � Firm maximize profits with respect to ζ δ P , , ij ij and γ 13 13 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

����������������       ζ + + δ γ ∞ 12 ∫ ∫      ( ) ( ) ( )  ε − ζ − δ + ε − ζ φ ε φ ε ε ε d d 12 12 23 23 23 23 12 12 23 12       ζ + δ ζ + sd + ζ + − δ ε      12 12 23 12  = + + + ζ + X a b P v T r   23 23 23 23 23 23    ζ + δ   ∞  12 ∫ ∫ ∫ ∞   ( ) ( ) ( ) ( )  + φ ε ε + φ ε ε ε − ζ φ ε ε   d d d         12 12 12 12 12 12 23 23 23 23 ij  ζ + sd        23  ζ + + δ γ 0 12 ( ) = + − + X a b P C Pr loose vEt 123 123 123 123 lf 123 = + ζ + + ζ + δ + δ + Et T T 123 12 12 23 23     ζ + + δ γ ζ + δ ∞ ∞ 12 12 ∫ ∫ ∫ ∫ ∫ ∞     ( ) ( ) ( ) ( ) ( ) ( ) ( ) ε − ζ − δ + ε − ζ φ ε φ ε ε ε + φ ε ε + φ ε ε ε − ζ φ ε ε d d d d d    12 12 23 23 23 23 12 12 23 12 12 12 12 12 12 12 23 23 23 23 ij  ζ + sd + r   23 ζ + δ ζ + ζ + − δ ε + sd 0 ζ + + δ γ   12 12 23 12 12   ( ) ( ) + −Φ ζ + δ + γ Ewt 1   12 12 14 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

���� Direct Flights Toulouse-Paris Paris-Nice Total passengers 177414 166831 Total number of flights 1432 1228 Average Passengers per 123.9 135.9 flight Travel time (minutes) Travel time (minutes) 80 80 85 85 Frequencies a 23.5 20.1 Airplane b A320 A320 Capacity c 161.9 168.1 Average occupation 76.5% 80.8% a Average frequency of flights per day; b Most frequent plane; c Average capacity of the used planes on the route 15 15 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

#������ � Calibration ( ) v ∈ 0.5,0.9 ( ) vr ∈ 0.84,0.93 ( ) r ∈ 0.95,1.96 � Optimal delays and optimal buffer time � Buffer decreases more than 50% � Extra delays decreases and dissapears in most of the cases 16 16 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011

����������� � Under the assumptions of linear demand, monopoly and same value of time for all the passengers we obtain that the buffer time as well as extra delays introduced by the airline should decrease � The introduction of compensation for long delays lead airlines to increase their prices. Overall effect over airlines to increase their prices. Overall effect over welfare is always negative 17 17 Miguel Urdanoz m.urdanoz@esc-toulouse.fr SESAR 2011