Organization of Multiple Airports in a Metropolitan Area

(1)

1

Organization of Multiple Airports in a Metropolitan Area

Se-il Mun and Yusuke Teraji Kyoto University

Full paper is downloadable at

http://www.econ.kyoto-u.ac.jp/~mun/papers/munAP081109.pdf

(2)

2

Multiple airports in the same area

in Osaka, Tokyo, New York, London,…

In most cases, each airport provides different

types of service.

(3)

3

Example: Osaka

http://www.sekaichizu.jp/atlas/japan/prefecture/map_p/p_osaka.html

①

②

③

11km

36km 26km

Itami: (D)

Kansai: (I, D) Kobe: (D)

CBD

(4)

4

Allocation of services

• In Osaka

Kansai: International and Domestic Itami: Domestic

• Alternative allocation Kansai: Domestic

Itami: International and Domestic

(5)

5

LaGuardia: (D)

JFK: (I, D) Newark: (I, D)

Example: New York City

15km

24km 25km

CBD

(6)

6

Example: London

①

②

Gatwick: (I, D) Heathrow: (I, D)

24km

45km

③

Stansted: (I, D) 55km

CBD

(7)

7

• In New York and Osaka

Airport closer to the CBD provides only domestic flights

Airport located at periphery provides int’l and domestic flights

• In London

All airports provide int’l and domestic flights

Locations of airports

Is there any relationship b/w

allocation of services and the airport location?

(8)

8

• In New York

All airports are operated by a single public authority.

• In Osaka

Each airport is operated by an independent authority.

• In London

All airports are operated by a single private authority.

Types of airport operation

Is there any relationship b/w

allocation of services and the airport operation?

(9)

9

This paper

• The model

- Two airports in the same area,

- Two types of service (International and domestic) - Users, carriers, airport operators

• Compare the alternative types of operation - Separate operation (PP):

Two private firms operate two airports separately - Integrated operation (M);

A single private firm operates two airports - Public operation (G):

Government operates two airports

(10)

10

Literature

- Pricing and investment for an airport Oum et al (1990), Brueckner (2002),

⇒ Focusing on a single airport - Multiple airports

Pels, et al (2000), De Borger and Van Dender (2006) Basso and Zhang (2006)

⇒ Focusing on a single type of service No studies on the allocation of services

(11)

11

The Model

• Setting

Within each region, the population is uniformly distributed.

A linear economy consisted from two regions: City and Hinterland

–b The City b

Hinterland Hinterland

ρC

Density:

ρH

(12)

12

• Location of airports

–b

The City

Hinterland Hinterland

x₁ Airport 1

b=x₂ Airport 2

– Each airport may provide international (I), domestic (D), or both (ID) flights.

– Only airport 1 is congestible

x

(13)

13

• The allocation of services between two airports aj: the service provided at airport j

: the allocation of services

(

a a1, 2

)

Rules of Notations:

(ID, ID)

Airport 1

Int’l &

Domestic

Airport 2

Int’l &

Domestic

(D, ID) ^Domestic Int’l &

Domestic

(I, ID) ^Int’l Int’l &

Domestic Int’l &

Domestic

(N, ID)

(14)

14

• Agents Users

choosing

- whether or not using the service - which airport to use

Carriers

determining

- the number of flights (=frequency) at airports Airport operators

setting

- airport charges, the service at its airport, or both

(15)

15

• The sequence of the decisions

1. Type of service (Airport Operators) N, I, D, or ID

2. Airport charges (Airport Operators) 3. Number of flights (Carriers)

4. Trip demand (Users)

(16)

16

• User cost

If an individual located at x uses the service S

provided at airport j

( )

4

S S

j j S j

j

C x t x x vh P

= − + F +

Fare

Access Cost

t :Traveling cost per a distance xj:Location of airport j

Scheduling Cost

v :Value of waiting time

S

Fj : Frequency of S at airport j h :Operating hours of the airport

(17)

17

• Demand

Number of trips for service S/ person:

inelastic and constant, d^S An individual uses service S at airport j if:

i) C^Sj

( )

x ≤ Ci^S

( )

x ^{for 1, 2,}j = i ≠ j^, ii) C^Sj

( )

x ≤ C^S ^{for 1, 2.}j =

Reservation price of service S

ASSUMPTIONS

i) C^D < C^I ii) d ^D > d^I

(18)

18

• Carriers

– Two carriers in each service market (k=1, 2)

( )

Sk S S S Sk

j Pj mj rj f j

π = σ − −

Marginal cost Airport charge

# of Flights

S

Pj

σ

:Fare

:# of seats

Revenue per a flight

– Profit of carrier k providing service S

from airport j – The competition of carriers: Cournot type

(19)

19

1 , 1

S S S

S I D

m = ω + c

∑

₌ f

2

S S

m =ω Operating Cost

Congestion Cost

S

mj differs in two airports:

– Marginal cost

(20)

20

• Airports

– Two types of airport operators:

Private firm and Government

– Airport operator determines the service provided at its airport, airport charges, or both.

(21)

21

– Private operation:

maximizing the revenue from airport charge

(1) Separate Operation: (PP) (2) Integrate Operation: (M)

maximizing the social surplus.

– Public operation: (G)

(22)

22

Variable Airport Charges and Service Choices

Alternative Regimes:

Airport 1 Airport 2

Regime PP

Regime M ^Private

Private Private

Regime G ^Public

- The service provided at each airport, - Airport charges,

aj S

rj

Operator chooses

(23)

23

Simulation

Parameters based on data in Osaka, Japan City Size: b=50 (km)

Population 16.4 million =>

Reservation prices so that passengers computed by

model best fit the observed data in Osaka, Kansai Airports:

164 thousand / km2

ρC = CS

Airport 2 (Kansai) Airport 1

(Osaka) Airport 2

(Kansai) Airport 1

(Osaka)

2089 9742

5596 -

Passengers in 2004

6824 8235

5583 -

Model

Domestic International

(24)

24

Regime PP

Each airport is operated by a private firm.

Airport 1 Airport 2 Private Private

The objective of airport j:

( ) ⁽

¹ ²

⁾

,

; , ^S ^Sk ; ,

j j i j j

S k

R ^{r , r}^j ⁱ a a =

∑

r f ^{r , r}¹ ² a a

(25)

25

Regime PP

- Airport charges

(

^,

)

^max

(

^; ^,

)

j j i j j i

R a a = R a a

j

* j i

r r , r for 1, 2, .j = i ≠ j

- Service choice

( ) ( )

(

^* ^, ^*

) (

^, ^*

^{( )} )

j j i j j i

R a PP a PP ≥ R a a PP

for 1, 2, .j = i ≠ j

The Equilibrium Allocation:

(

^{a PP a PP}¹^*

( ) ( )

^, ²^*

)

(26)

26

Regime PP

Airport 2

–50 –25 0 25 50

x₁

Payoff Matrix (unit: billion Yen) 2

1

I D ID

I 65, 61 188, 259 65, 320

D 243, 189 84, 89 84, 278

ID 301, 65 272, 89 148, 146

(ID, ID)

(27)

27

Regime PP

Airport 2

–50 –25 0 25 50

x₁

1

I D ID

I 64, 63 188, 259 64, 306

D 243, 189 67, 53 67, 252

ID 306, 63 255, 53 131, 88

(ID, ID)

(28)

28

Regime M

Two airports are operated by a private firm.

Airport 1 Airport 2 Private

- Airport charges

(

₁ ₂

) (

₁ ₂

)

, 1,2 1,2

, max _j^S _j^Sk ; , .

S I D j k

R a a r f a a

= = =

=

∑ ∑ ∑

1 2

r ,r r , r

- Service choice

( ) ( )

( ) ⁽ ⁾

1 2

* *

1 2 1 2

, arg max, , .

a M a M = a a R a a

(29)

29

Regime M

( )

*

aj M : Service provided at Airport j

Airport 2

–50 –25 0 25 50

x₁

The Allocation with Variable Airport Charges:

( ) ( )

(

^a¹^* ^M ^,^a²^* ^M

)

(ID, ID) (D, ID) (N, ID)

(30)

30

Regime M

Airport 2

–50 –25 0 25 50

x₁

1

I D ID

I 202 447 461

D 424 274 463

ID 431 514 528

(31)

31

Regime M

Airport 2

–50 –25 0 25 50

x₁

1

I D ID

I 193 437 452

D 447 264 453

ID 435 442 450

(32)

Regime G

Two airports are operated by the government.

Airport 1 Airport 2 Public

Social Surplus:

(

1 2

)

2 2 2

1 1 1

, , ,

; ,

^S_j ^Sk ^S_j

j k j

S I D S I D S I D

SW a a

CS π R

= = =

=

∑ ∑

+

∑ ∑

+

∑ ∑

1 2

r ,r

Profit of carriers

Airport charge revenue

Consumer Surplus: ^CS^S^j ⁼

_∫

^q^S

( )

^{x C}

(

^S ⁻^C^S^j

( )

^x

)

^dx

(33)

33

Regime G

- Airport charges

(

1, 2

)

max

(

; ,1 2

)

. SW a a = SW a a

1 2

r ,r r , r

- Service choice

( ) ( )

( ) ⁽ ⁾

1 2

* *

1 2 1 2

, arg max, , .

a G a G = a a SW a a

(34)

34

Regime G

( )

*

a Gj : Service provided at Airport j

Airport 2

–50 –25 0 25 50

x₁

( ) ( )

(

^{a G a G}¹^* ^, ²^*

)

(35)

35

Regime G

Airport 2

–50 –25 0 25 50

x₁

Comparison of Social Surplus: (UNIT: billion yen) Social

Benefit

Access Cost

Scheduling Cost

Congestion Cost

TOTAL

(ID, ID) 1704 575 20 28 1081

(D, ID) 1690 580 17 16 1077

(ID, D) 1648 545 17 44 1042

(N, ID) 1669 633 10 0 1026

(36)

36

Regime G

Airport 2

–50 –25 0 25 50

x₁

Comparison of Social Surplus: (UNIT: billion yen) Social

Benefit

Access Cost

Scheduling Cost

Congestion Cost

TOTAL

(ID, ID) 1645 575 20 43 1007

(D, ID) 1596 527 17 24 1028

(ID, D) 1607 537 17 62 991

(N, ID) 1669 633 10 0 1026

(37)

37

Comparison

Airport 2

–50 –25 0 25 50

x₁

PP

M

G

(ID, ID)

(38)

38

Comparison

Difference in Domain of (ID, ID): PP > M > G

– Excess use of airport 1 under the private operations – The integrated operation (M) allocates the services

more efficiently than the separate operation (PP)

(39)

39

Regime PP v. s. M

Airport 2

–50 –25 0 25 50

x₁ PP

M

(ID, ID)

(40)

40

Regime PP v. s. M

Regime PP: Payoff Matrix (unit: billion Yen) 2

1

I D ID

I 64, 63 188, 259 64, 306

D 243, 189 67, 53 67, 252

ID 306, 63 255, 53 131, 88

Airport 2

–50 –25 0 25 50

x₁ PP

M

(ID, ID)

(41)

41

Regime PP v. s. M

Regime M: Payoff Matrix (unit: billion Yen) 2

1

I D ID

I 193 437 452

D 447 264 453

ID 435 442 450

Airport 2

–50 –25 0 25 50

x₁ PP

M

(ID, ID)

(42)

42

Regime PP v. s. M

S

Rj

∆ : Change in the revenue from service S at airport j Regime PP

(ID, ID) (D, ID)

Regime M

( ) ( )

1 , 1 , 1^I 1^D 0

R D ID − R ID ID = ∆R + ∆R <

( – ) ( + )

( ) ( )

2 2

1 1 2 2

1 1

, , ^I ^D + ^I ^D 0

j j

R D ID R ID ID ∆R ∆R ∆R ∆R

= =

− = + + >

∑ ∑

( – ) ( + )

- Airport 1 loses its service I market.

- Airport 2 receives the gain.

PP: The operator of airport 1 neglects.

(43)

43

Regime M v. s. G

Airport 2

–50 –25 0 25 50

x₁ G

(ID, ID) (D, ID) (N, ID) (ID, ID) (D, ID)

M (N, ID)

(44)

44

Regime M v. s. G

(ID, ID) (D, ID)

Regime M

Regime G

(

^,

) (

^,

)

⁰

SW D ID − SW ID ID = ∆CS + ∆PS + ∆R >

( – ) ( + )

Increase in Access Cost

Decrease in Social Benefit < Decrease in Scheduling Cost Decrease in Congestion Cost

( ) ( )

2 2

1 1 2 2

1 1

, , ^I ^D + ^I ^D 0

j j

R D ID R ID ID R R R R

= =

− = + + <

∑ ∑

^∆ ^∆ ^∆ ^∆

( – ) ( + )

(45)

45

Comparison

Airport 2

–50 –25 0 25 50

x₁

PP M G

(ID, ID)

(ID, ID) (D, ID) (N, ID) (ID, ID) (D, ID) (N, ID)

Service I Service D

Airport 1 Airport 2 Airport 1 Airport 2

PP (ID, ID) 7418 6959 3746 3320

M (D, ID) - 17630 5102 5488

G (D, ID) - –17630 –8085 –8245

Comparison of Airport Charges (unit: thousand Yen)

(46)

46

Comparison

Airport 2

–50 –25 0 25 50

x₁

PP M G

(ID, ID)

(ID, ID) (D, ID) (N, ID) (ID, ID) (D, ID) (N, ID)

Comparison of Social Surplus (unit: billion Yen) Social Surplus

PP (ID, ID) 797

M (D, ID) 744

G (D, ID) 1058

(47)

47

Comparison

– The integrated operation (M) allocates the services more efficiently than the separate operation (PP).

– BUT the integrated operation (M) charges the services more than the separate operation (PP).

Welfare Ranking:

PP ; M

Efficiency losses of pricing: more significant

(48)

48

Surplus-Maximizing Allocation

The Surplus-Maximizing Allocation under Regime T:

- Airport charges, ^r_j^S

determined by the operator

aj

- The service provided at each airport,

set to maximize the social surplus

( ) ( )

( ) ⁽ ⁾

1 2

1 2 1 2

, arg max, , ;

O O

a T a T = a a SW a a T

(49)

49

Surplus-Maximizing Allocation

Airport 2

–50 –25 0 25 50

x₁

PP

M

G

(ID, ID)

(50)

50

Comparison: Variable

Airport 2

–50 –25 0 25 50

x₁

PP

M

G

(ID, ID)

(51)

51

Surplus-Maximizing Allocation

By comparing with the case of variable airport charges:

1) The allocation - Regime PP:

the same as the variable case - Regime M:

the difference is quite small 2) Airport charges:

determined by the operator Welfare Ranking:

the same as the case of variable airport charges

(52)

52

Allocation under Parametric Airport Charges

Assume that

airport charges are exogenously given.

(by regulation or by other reasons) Airport charges are set as (Kansai Airport):

1^I 2^I 1537.39 (thousand yen), r = r =

1^D 2^D 718.08 (thousand yen).

r = r =

Operators set the type of services at its airport.

(53)

53

Airport 2

–50 –25 0 25 50

x₁

Allocation under Parametric Airport Charges

PP

M

G (ID, ID) (D, ID)

(N, ID)

(54)

54

Allocation under Parametric Airport Charges

Airport 2

–50 –25 0 25 50

x₁

Social Surplus

PP (ID, ID) 889

M (D, ID) 889

G (D, ID) 893

Comparison of Social Surplus (unit: billion Yen) G

PP M

(ID, ID) (D, ID) (N, ID) (ID, ID) (D, ID) (N, ID) (ID, ID) (D, ID) (N, ID)

(55)

55

Allocation under Parametric Airport Charges

Welfare Ranking: PP = M Regimes PP and M:

The same allocation at all locations of airport 1

The choice of carriers affects the allocation Airport charges: exogenously fixed

Difference in Social Surplus among Allocations:

Relatively Small

(56)

56

Summary

(1) The allocation of services varies with the locations of airports in a metropolitan area.

(2) If the airport charges are variable, the private

operation induces the excessive number of services.

(3) Welfare gain from the regulation on the service choice is quantitatively small.

(57)

57

Topics for the Future Research

- Mixes of Private and Public Airports - Capacity Choice (Number of Runways) - Three Airports in a Same Area

(58)

58

(59)

59

• Parameters

Parameter Unit

153 (thousand yen) 41 (thousand yen) 0.17 (times/ year) 0.73 (times/ year)

3 (thousand yen/ hour) 5475 (hours/ year)

272 (seats) CI

CD

d I

d D

h v

σ

Parameter values: based on the observation

(60)

60

Parameter Unit

13522 (thousand yen) 2015 (thousand yen) 0.01 (thousand yen) 0.1 (thousand yen)

164 (thousand people/ km²) 26 (thousand people/ km²)

50 (km)

ρH^C

ρ

b c

t

ωI

ωD

(61)

61

Regime PP

( ) ( )

(

^a¹^* ^{PP a}^, ^*² ^PP

)

(ID, ID)

( )

*

aj PP : Service provided at Airport j

Airport 2

–50 –25 0 25 50

x₁

(62)

62

Regime PP

( ) ( )

(

¹^* ²^*

) (

¹

^{( )}

²

^{( )} )

Comparison of a PP a, PP and a^O PP a, ^O PP

(ID, ID)

Airport 2

–50 –25 0 25 50

x₁ (ID, ID)

( ) ( )

(^a¹^* ^{PP a}^, ^*² ^PP )

( ) ( )

(â¹Ô ^{PP a}^, ²Ô ^PP )

(63)

63

Regime M

( ) ( )

(

¹^* ²^*

) (

¹

^{( )}

²

^{( )} )

Comparison of a M ,a M and a^O M ,a^O M

Airport 2

–50 –25 0 25 50

x₁

( ) ( )

(^{a M}¹^* ^,^a^*² ^M )

( ) ( )

(â¹Ô ^M ^,â²Ô ^M )

(64)

64

Sensitivity Analysis

Case A: service I is charged higher than service D

1^I 2^I 718.08 (thousand yen), r = r =

1^D 2^D 1537.39 (thousand yen).

r = r =

Case B: airport 1 is charged higher than airport 2

1^I 1^D 867.60 (thousand yen), r = r =

2^I 2^D 718.08 (thousand yen).

r = r =

(65)

65

Airport 2

–50 –25 0 25 50

x₁

Case A

PP

M

G (ID, ID) (D, ID)

(N, ID)

(66)

66

Airport 2

–50 –25 0 25 50

x₁

Case B

PP

M

G (ID, ID) (D, ID)

(ID, D) (ID, N)

(N, ID)

(67)

67

Sensitivity Analysis

The design of airport charges affects

Welfare Ranking b/w PP and M Regime M:

Regimes PP and G:

Qualitatively similar to the Base case

The design of airport charges affects the allocation

(68)

68

Example: Seoul

CBD Gimpo: (D)

Incheon: (I, D)

52km

16km

(69)

69

Example: Paris

14km

Orly: (D) CBD

26km

Charles De Gaulle: (I, D)

(70)

70

Example: Tokyo

19km

Haneda: (D) CBD

66km

Narita: (I, D)

①

②

Organization of Multiple Airports in a Metropolitan Area