Airline hubbing—Some implications for airport economics.pdf

8/11/2019 Airline hubbing—Some implications for airport economics.pdf

http://slidepdf.com/reader/full/airline-hubbingsome-implications-for-airport-economicspdf 1/13

TWW~. RCS..A Vol. 19A, No. I. pp. 15-27. 1985

Printed in the U.S.A.

0191-2607185 f3.00 .lO

Pcrgamonresstd.

AIRLINE HUBBING-SOME IMPLICATIONS

FOR AIRPORT ECONOMICS

ADIB KANAFANI

Institute of Transportation Studies, University of California, Berkeley, CA 94720, U.S.A.

and

ATEF A. GHOBRIAL

Manalytics, Inc., San Francisco, CA, U.S.A.

(Received

April

1984)

Abstract-The phenomenon of airline hubbing has been on the increase in recent years. Hubbing arises when

airlines attempt to maintain high levels of aircraft utilization and to take advantage of scale economies. Passengers

also appear to benefit from hubbing in the form of increased frequency of service. The nature of traffic generated

at the hub airports implies some negative economic impacts which suggest that hub pricing should be considered

seriously. This study shows hubbing to be

“inelastic” to hub pricing and concludes that there are significant

potential benefits to the airports to be gained from some form of hub pricing.

INTRODUCTION

Of the 21 million passengers who enplaned at Atlanta’s

Hartfield Airport in 1979, 16 million were simply con-

necting between flights. Similarly, 11 of the 21 million

enplaned at Chicago’s

O’Hare Airport and 6 of the 15

million enplaned at Los Angeles International Airport

were connecting passengers. During the fourth quarter

of 1980, nearly 30% of all travelers between Tallahassee

and Miami chose connecting flights, despite the avail-

ability of 21 weekly non-stops on the route, paying an

average of 10 more than the non-stop fare and incurring

a journey time that was 50% longer.

These and many similar statistics describe a phenom-

enon that has become increasingly prevalent in the air

transportation system: airline hubbing. Hubbing occurs

when airlines concentrate flights at a few airports which

they use as collection-distribution centers for their pas-

sengers. Unlike the situation in the early days, when

airlines had to make technicaf refueling stops, today’s

hubbing is motivated by the economic advantages of

increased flight frequencies and by the economies of

operating larger aircraft. High frequencies of service and

larger aircraft would not normally be feasible if all city

pairs in a system were served by non-stop flights. By

consolidating passengers through a few selected airport

hubs, an airline takes advantage of the resulting higher

volumes by using large relatively efficient aircraft and

can raise the frequency of service it offers passengers to

offset the increased travel time occasioned by the need

to transfer.

Seen from the perspective of airport economics, airline

hubbing would appear both to pose problems and to offer

opportunities. The problems are not simply due to the

added traffic volumes generated, for traffic volumes are

not undesirable in an era of limited growth. The problems

are rather due to the nature of the traffic brought to the

airport. The opportunities arise from the realization that

both airlines and passengers appear to share in the eco-

nomic benefits of hubbing. The question then arises

whether the airports involved can also benefit. The pur-

pose of this paper is to explore some of the impacts of

hubbing on airport economics.

AIRLINE HUBBING

Airline hubbing is a phenomenon that can be consid-

ered on the rise in the U.S. air transportation system.

This is suggested in Table 1, which shows the evolution

of connecting passenger percentages at the eight top U.S.

airports. Casually, a look at any airline’s route map

would suggest that one or two airports serve as major

hubs and as collection-distribution centers for the dif-

ferent airlines: Dallas-Fort Worth for American, Chicago

O’Hare and Denver for United, Atlanta for Delta, and

so forth. Figuresl(a) and (b), comparing the regional air

networks of the southeastern United States in 1960 and

1970, provides a strong indication of the importance of

modem hubbing. Atlanta is the region’s major hub, and

one that must be used for travel between many of its

city pairs. More recent networks, such as the one in 1982

shown in Fig. 2, suggest that the hubbing tendency has

continued to flourish even after deregulation of the air-

lines in 1978. Despite the establishment of numerous

direct links between city pairs, the network continues to

be strongly concentrated at its major hub, and many

smaller cities are still connected to the air transportation

system only by a link to that hub. The historic evolution

of hubbing and the role aircraft technology played in it

have been studied by Kanafani (1980) and Kanafani and

Ghobrial (1982). Before moving on to the examination

of airport impacts, a brief discussion of the forces that

motivate hubbing is in order.

WHY Do AIRLINES HUB?

If it were possible to operate economically with single-

seat aircraft, then all passengers could be served directly

15



16

A. KANAFANIand A. A. GHOBRIAL

Table 1. Airport Enplanements and Percentage of

Connecting Passengers (Enphment in thowmds)

an Francisco

allas Fort Wort

I

19

Passenger

nphlement

13,201

6.379

7,662

5.257

5,192

4,549

4,139

3.175

ercentag

xmectinl

47

71

33

32

53

29

22

21

19

Passenger

nplanement

16,252

12.507

9.084

6,063

7,296

6.456

4,953

3,343

ercentag

xlnectin

52

76

31

31

59

37

25

23

c

-

19

Passenger

nplanement

21.119

20,797

15.901

10.763

11.419

7.564

7,093

4.526

:rcentag~

xmectirq

51

76

40

41

55

35

26

22

1) U.S. Civil Aeronautics Board

Domestic origin destination survey ol airline

passenger tradlc Table 11)

2) U.S. Federal Aviation Administration Airport

activitv statis tics of certided route air

carrieis Table 3)

between their points of origin and destination, and at the

desired time. There would be no need for passengers to

transfer, and the phenomenon of hubbing would not arise.

Major hubs would, of course, still emerge at the larger

cities, since more people would fly to and from them

than to and from the smaller cities; but there would be

no connecting traffic.

As soon as aircraft size begins to offer economies of

scale and to dictate a schedule, the situation changes

fundamentally. The large traffic volumes to and from

the larger cities in a region encourage the use of larger

aircraft. However, in the interest of maintaining the level

of service, an airline wants to keep some lower bound

on schedule frequency. Thus, the need to “fill” aircraft

that are flying to these cities arises and causes the airline

to hub by reducing the direct service between smaller

cities and by offering connecting flights at the hub. This

consolidation of passengers in links to and from the hub

allows the airline to take advantage of the economies of

aircraft size. Consolidation also allows both the airline

and its passengers to take advantage of the economies

of increased schedule frequency: The airline, by offering

a high frequency, ensures itself a prominent share of the

market; and the passengers, by using connecting ser-

vices, enjoy the convenience of frequent flights.

The forces that encourage airlines hubbing are pri-

marily a function of aircraft technology, influenced, of

course, by market considerations such as frequency com-

petition and airline presence at major airports. There are

no particular scale economies at the airports themselves:

If frequency of service were not such an attractive level-

of-service attribute, and if it were not for the need to fill

0

MAJOR

HUB

MEDIUM HUD

SMALLHUD

NONHUD

MO

b

TLH

74x

PNS

(A)

1960

Fig. 1. Regional air networks.



Airline hubbing and airline economics

17

BOSTON

Fig. 2. Air networks, 1982.

the larger aircraft and to take advantage of the resulting

lower passenger-kilometer costs, there would be no strong

incentive for the airline to consolidate its operations at

an airport. Much of the available evidence concerning

station costs, such as that shown in Fig. 3, suggests that

average costs per passenger do not decline with volume

at an airport. Nor do landing fees, shown in Fig. 4 for

a sample of 22 U.S. airports, seem to drop as the volume

at an airport increases. The economies of scale that en-

courage hubbing are link features and not node features

in the air network, and result, as mentioned before, from

the characteristics of larger aircraft and from the increas-

ing returns in terms of market share that can be achieved

by increasing frequency share in a market with com-

petitors.

Figure 5 shows a comparison of market and frequency

share for a selection of markets in the southeastern United

States in 1980. This comparison reaffiis the well-known

S-shaped relation and provides an additional explanation

of the factors that underlie airline hubbing. Some of the

figures quoted at the outset of this essay support this and

suggest that some passengers are indeed willing to incur

additional travel time and, in some cases, cost in order

to take advantage of higher schedule frequency.

SOME IMPLICATIONS FOR AIRPORT ECONOMICS

Airline hubbing affects airport economics in many

ways. At a number of airports, connecting traffic oc-

casioned by hubbing is predominant. This alters fun-

damentally the economic role of the airport within the

region it serves. A major connecting airport represents

more of an airline collection-distribution facility than a

transportation facility for a region or city. The impli-

cations of this for airport economic policy are important

and will be the subject of further discussion below. In-

creased connecting traffic brought about by hubbing has

characteristics that differentiate it from other types of

traffic as far as airport economic impacts are concerned.

Connecting traffic is characterized by a banked sched-

ule usually designed to allow a group of incoming flights

to arrive within a relatively short period of time and to

be followed, after an appropriate connecting time inter-

val, by a group of departing flights. This normally will

result in increased peaking patterns. These peaks occur

over relatively short periods of time, say on the order

of 1 hr, and are often not captured by conventional typ

ical peak-hour statistics. The effects of this increased

peaking are to demand more capacity per unit of traffic

and consequently to place an added burden on airport

facility development. In short-haul systems, the con-

necting time is rather short, typically no more than an

hour. This implies that passenger flow is primarily a

matter of processing and movement from one gate to

another. Consequently, the revenue generation potential

from airport concessions is decreased. Taken together

with the absence of revenue generated from access sys-

tem use (e.g. parking revenues), this means that on a

per-passenger basis the airport is not able to generate the

same amount of revenue as for originating traffic. Yet



18

A. KANAFANI and A. A. GHOBRML

UNITED AIRLINES

0

1979)

0

‘0

3uu

IUUU

13uu

PASSENGER ENPLANEMENTS 1000)

Fig. 3. Airline station costs and airport traffic. (Source: CAB Form 41.)

the fees the airport receives for the use of its facilities

are the same, whether they be landing fees or space

leases.

Another, perhaps more important, economic impact

of hubbing is that it makes the hub airport more de-

pendent on airline networking considerations and less

dependent on the economic demand for travel within the

region it serves. The airport becomes more affected by

airline entry and exit decisions, which adds a dimension

I-

)-

I-

3-4

I-

L

of uncertainty to the airport’s financial planning. When

an airline decides to make an airport its transfer hub (as

recently occurred, for example, in Kansas City), a sud-

den surge in traffic and facility demands results. When

the opposite occurs, it is potentially more threatening,

as a sudden disappearance of a significant segment of

an airport’s traffic could result in loss of sponsorship

and revenues. The resulting threat to airport financial

planning is not difficult to comprehend, especially as an

B

60

IZU IUU L4U 255

AIRCRAFT OPERATIONS 1000)

Fig. 4. Landing fees and airport traffic.




19

100

8

20

2 AIRLINES

0 3 AIRLINES

I 1 1 I

20

40 60

80

IC

FREQUENCY SHARE

Fig. 5. Airline frequency and market share.

airport’s ability to obtain credit could be compromised.

An airport that is predominantly a connecting hub does

not contribute as much to its local economy as one that

serves predominantly originating traffic. The exigencies

of financial independence become more compelling when

the case for social benefits in the host region is thus

weakened. A connecting hub airport must find ways to

rationalize its function economically. If there are benefits

to be gained from hubbing, might the airport not join

the passengers and the airlines in sharing them? Could

a major transfer hub establish a pricing mechanism that

would achieve this? If we take a more global perspective

than that of the airport concerned, the question also arises

regarding the extent to which hubbing results in hidden

subsidies to users. Given that most of the heavy federal

investments in, for example, the air traffic control sys-

tem, are aimed at alleviating congestion at major hubs,

of which a good number are transfer hubs, it can be

argued that airlines paying average cost at uncongested

airports may be subsidizing those that create the conges-

tion partly by their network concentration. A classical

argument of marginal cost pricing would follow directly

if this suggestion were pursued. We will look at the effect

of such a pricing strategy on the hubbing phenomenon

and, consequently, on airport economics.

A HUBBING CASE STUDY

In an effort to explore the effects of pricing on network

hubbing and airport economics, a case study conducted

for the southeastern United States region is shown in

Fig. 2. The study focused on part of this region, in-

cluding about 25 cities for which Atlanta is the major

connecting hub and for which Nashville, Tennessee;

Charlotte, North Carolina; Birmingham, Alabama; and

Jacksonville, Florida, appear to be secondary hubs. A

larger part of the air network was included, as shown in

the figure, to deal with the problem of study area delin-

eation. Traffic between the major hubs outside the region

of immediate concern, such as between Miami and

Washington, or New Orleans and New York, is not in-

cluded in the study since this traffic flows predominantly

(upward of 95%) on non-stop routes. For the remaining

“regional” network, an equilibrium model was con-

structed following the logic outlined in Fig. 6.

The basic characteristic of this model is that it assigns

origin-destination

flows

to the service network using a

calibrated route choice submodel. Using the resulting

link flows, the model then assigns frequency of service

on each link so as to maximize the level of service

without going below the break-even load factor on that

link. The process is iterated with the route choice ad-

justed according to relative levels of service on alter-

native routes until an equilibrium is reached. Entering

into the computation of the break-even load factor is the

cost function for a given aircraft technology (in this case

a hypothetical hybrid of DC9 and B727) and whatever

pricing penalty is imposed at the hub airport (Atlanta).

The evaluation of system performance is then based on

airline profitability (notice that there is no loss, since all

links at least meet the break-even load factor interim),

level of service (frequency, delay and fares) and hub



20

A.

KANAFANI and A. A. GHOBRIAL

E[

PRIMARY

NETWORK 1

O-D DEMAND

CA

INK LOW

I

EVALUATION OF

SYSTEM PERFORMANCE

I

Fig. 6. Network equilibrium model.

airport revenues as influenced by the pricing level pos-

tulated.

The route choice model

A sample of 19 city pairs in the study region was

selected for calibration of the route choice model. These

city pairs, shown in Table 2, are typically connected by

non-stop service, as well as connecting service through

Atlanta. Observations of city-pair traffic on each of the

routes connecting them were made for the second quarter

of 1980 together with observations of route schedule

frequencies, airfares, aircraft types and the nature of the

connection (whether a non-stop, direct or indirect con-

nection). A multinomial route choice model of the logit

type was then constructed and calibrated with these ob-

servations. The choice function was constructed as fol-

lows:

V(r,j) = e,T,j + a ,, + afTi

+

aaDd + a,C,,

1)

where T, is the travel time on route r for city pair j; F

is the daily frequency of service on a route;fis the one-

way economy class airfare on a route;

D

is a dummy

variable for aircraft size taking on the following values:

0 for N I 30, 0.5 for 30 < N < 50, 1.0 for N 2 50,

where N is the number of seats); and C is a dummy

variable for the connectivity service taking on the fol-

lowing values: 0 for non-stop service, 0.5 for online

service with stops, 1.0 for indirect connecting service.

The route choice probability for traffic between a given

city pair is then given by

exp Mr. ill

P(r9’)

2

exp

[V r, j)]

The results of the calibration using the maximum like-

lihood method are shown in Table 3. It is interesting to

note the high level of significance of the parameter es-

timates. In itself the model shows some interesting facts

about the observed route choice. The values of both




Table 2. City-Pair used in Calibrating the Route Choice Model

21

dummy variable parameters being both rather significant

and of a similar magnitude suggests that aircraft type

and directness of service are both equally important.

People do appear to prefer larger aircraft. The difference

between a non-stop and connecting flights appears to be

worth about 2 hr of travel time, which is not surprising.

The marginal rates of substitution between time, fre-

quency and airfare suggest a revealed valuation of travel

time at about 3O/hr, and a valuation of an additional

daily flight at about 1 llhr at the margin. These are

significant amounts in short-haul systems where the air-

fares are on the order of 50- 100.

Airline operating model

In

order to evaluate network performance from the

perspectives of operating costs and level of service, a

Table 3. Estimated Coefficients of the Route Choice Model

Estimated Coefficient

T-V C

Time (hour)

Daily Frequency

Fare (dollar)

Conneztivity Pattern

-14.77

38.21

-6.7

15.4

14.0

Chi-Square Value 3114

Log-Likelihood Value -34.611

simple model of airline operation over the study network

was constructed. The model computes airline operating

costs and flight times for each flight segment. For the

application in this case study, an aircraft of the DC9-

30/B727-200 type was assumed. Operating cost data

over stage lengths similar to those in the study region

are estimated at 2194 per block hour, and the indirect

cost was estimated at 12 per passenger.

The travel time function for the aircraft in question is:

T = 0.2353 + 0.002R.

(3)

where

T

is the block time in hours and

R

is the link stage

length in miles.

In evaluating the operation over a link connected with

the hub airport, it was necessary to construct a delay

function to estimate the effect of congestion at the hub

on link travel time and route choice. Not knowing the

magnitude of the flow at the hub in total prevents the

exact estimation of delay as caused by the connecting

flights. Consequently, it was necessary to make the as-

sumption that the connecting airport is operating at a

level where the delay function is cubic in traffic flow.

This results in the following marginal delay function:

a Zn (d) = 3 a Zn (V),

(4)

where

d

is the average delay at the connecting hub airport

and

V

is the daily volume at the airport. Changes in the



A. KANAFANI and

volume caused by adjustments in the network will result

in changes in the average delay level and the travel time

on links connected to the hub. In order to complete the

level-of-service characterization, a submodel was nec-

essary to calculate passenger travel time. This differs

from the airline time by the component usually referred

to as frequency delay. Frequency delay is inversely pro-

portional to service frequency. The assumption made

here is that the average headway h in between flights on

a link is related to the service frequency

F

on that link

by

The expected average frequency delay associated with

that frequency is a quarter of the average headway (as

opposed to the one-half usually assumed with random

arrival of passengers taking flights). Consequently, the

frequency delay L on a link is given by

The total travel time for a passenger on a link is then

made up of the aircraft block time as given in eqn 3, to

which is added the frequency schedule delay from eqn

6 and, when a transfer is involved, an additional hour.

The effect on network structure of congestion pricing

at the connecting hub is explored by parametrically vary-

ing a penalty level assumed charged at the hub. The

actual landing fees at Atlanta during the study year were

DALLAS-

ORT M

rORTH// kj

_ _- ___

NEW YORIL 1

A. A. GHOBRML

at a rate of 37 / 1000 lb of landing weight. For an aircraft

of the type considered here this would amount to about

50 per landing. This landing fee structure results in a

total annual revenue of about 12 million. In applying

a congestion toll at the airport, the total operating cost

on a link would increase accordingly. This increase could

be. converted to equivalent minutes of delay using the

operating cost functions included in the submodel. A

penalty of about 360 would be equivalent to an addi-

tional delay of 10 min per aircraft. With this type of

conversion, it becomes possible to assess the impact of

either a congestion toll at the airport or an actual increase

in delay that might result from a capacity shortfall. The

effect on network structure is analyzed by varying this

penalty within the range of 0- 1800 per landing, or

equivalently within the delay range of O-SO min.

Case study resul ts

Two strategies are postulated fo represent airline re-

sponse to the additional penalty at the hub airport. One

is to absorb the additional operating cost and recover it

by raising the break-even load factor on the affected

links. This means a reduction in schedule frequency on

these links. The other strategy is to maintain the same

break-even load factor by passing the additional cost

directly to passengers in the form of increased airfares

on the affected links. Passengers ultimately pay the pen-

alty in both cases: indirectly in the first through increased

frequency delay, and directly in the second through

higher airfares. Fewer people will select the connecting

flight through the hub, and the network will be decon-

centrated as a result. The revenues from the hubbing

BOSTON

NEW ORLEANS

--- NEW DIRECT LINK

(A) 360

Fig. 7(A). Air networks with hub penalties.

.PHIA




23

penalties represent added income for the hub airport-

in relative terms a rather significant amount, as the results

indicate.

The parametric study of the effect of pricing on net-

work structure suggests, in both response cases, that

hubbing will continue to predominate even with penalties

as high as 1500 per landing. These results appear to be

quite similar for the case of airfare being increased and

the case of the break-even load factor being modified to

absorb the extra cost. Consequently, we shall report only

on the latter case.

The effect on network structure of pricing at the major

hub is expected to be the introduction of direct services

between cities, thereby relieving the pressure of con-

necting flights. Figures 7(a)-(d) shows these effects

graphically. In the first case, a penalty of 360, which

is equivalent to an additional delay at the hub of 10 min,

is applied. This results in the addition of three direct

links between cities, as shown in Fig. 7(a). When the

penalty is tripled to 1080, or 30 min of delay, three

additional direct links are established. Finally, with a

penalty of 1800, equivalent to a delay of 50 min, the

totai number of new direct links in the study network

reaches seven. The effect on the traffic at the major hub

of this admittedly limited network decentralization is

depicted in Fig. 8. The number of annual operations at

Atlanta declines from 450,000 to 400,000 with a penalty

of 360, to 340,000 with a penalty of 1080 and to

280,000 with a penalty of 1800. These figures suggest

an airline demand for flights to the hub with a price

elasticity in the range of - 0.148 to - 0.38. Already the

implication for airport economics, and the opportunities

to raise airport revenues while simultaneously maintain-

ing airline profitability and level of service. are clear.

Before exploring these further, a look at the effect of

this dehubbing on level of service is in order.

Figure 9 shows the changes in average load factor that

take place as the hub penalty is raised. An increase in

the penalty level from 0 to 1800 brings about an in-

crease in system load factor from 65 to 82%. While the

upper level of 82% may be considered a deterioration in

level of service, it is clear that there is a point somewhere

in this range, say about halfway, where an increased

load factor is tolerable, especially if this increase is con-

comitant with a reduction in congestion delay. Figure

10 shows the delay trends caused by the pricing system.

Congestion delay at the hub is seen to decline from about

30 min to about 15 min within the pricing range of O-

1800. Figure 11, showing the resulting airline operating

costs, suggests that the savings in operating costs oc-

casioned by the reduction in congestion are more than

adequate to cover the penalty charged. The total cost

does not increase.

Returning to Fig. 10, we see that frequency delay

increased because of the reduction in flight frequency at

the hub. This increase is more than the reduction in

congestion delay. Consequently, there is a deterioration

of the level of service. However, given the manner in

which frequency delay is defined and calculated, the

resulting increase from 1.3 to 1.4 hr may not represent

a serious deterioration of level of service. From all these

results it would appear that imposing a hub penalty will

not alter the network structure significantly. Airline op-

erating costs will not change appreciably either. Passen-

BOSTON

Fig. 7(B). Air networks with hub penalties.



24

A. KANAFANIand A. A. GHOBRIAL

Fig. 7(C). Air networks with hub penalties.

BOSTON

-_--____

CLEVELAND

m

A I

PI I

ID) 1800

Fig. 7(D). Air networks with hub penalties.




25

600 -

360

720

1080

PENALTY

LEVEL

Fig. 8. Effect of hub penalty on airport traffic.

gers will have to contend with reduced service frequen-

cies and increased load factors; but, even then, connecting

at a major hub will continue to be a preferred option for

most regional traffic.

Turning to the hub airport itself, we see from Fig. 12

that the increase in penalty will generate significant rev-

enuez. Considering the revenues at a penalty level of

1080, for example, we find a revenue potential of about

200 million annually compared with the actual landing

fee revenue of about 13 million during the study year

1980. Again, there is a great potential for generating

revenues which, if judiciously used to implement delay

reduction measures, could further increase overall sys-

tem efficiency.

SUMMARY AND CONCLUSIONS

The forces that encourage airline hubbing are com-

pelling and result from two major characteristics of air

0.6

1440

I000

transportation. The first is the indivisibility of aircraft

size and the scale economy it creates, a characteristic

common to all scheduled transport systems. The second

is the preference of travelers for high levels of schedule

frequency. Airlines are able to achieve relatively high

load factors by consolidating flights at major connecting

hubs. Passengers are willing to incur additional travel

time, and sometimes additional travel costs, in order to

take advantage of the higher frequencies available through

these transfer hubs.

The case study conducted using the southeastern United

States suggests that congestion pricing, or the hubbing

penalty imposed at the connecting airports, will not sig-

nificantly discourage hubbing. In general it would appear

that such penalties can be absorbed by raising the break-

even load factors, thereby causing the frequency of

service to decline. In other words, these penalties are

absorbed primarily by the passengers in the form of

increased frequency delay.

360 720 1080 1440

PENALTY LEVEL

Fig. 9. Effect of hub penalty on average load factor.

1800



26

A. KANAFANI and

A

A. GHOBRIAL

= 0.8

c3

b.

--__

0.4

- ---__ _

----_

CONGESTION DELAY

----..-.._--__

-----________

I I , I 1 I I I 1

360 720 1080 1440 1800

PENALTY LEVEL

Fig. 10. Effect of hub penalty on delays.

There are important implications for airport econom-

ics. First, it is clear that there is significant potential for

generating increased airport revenues. These revenues

can play an important role in the implementation of de-

lay-reducing measures which would increase the overall

efficiency of the system. Revenues can also be seen as

payment by the users of large congested hub airports for

the externalities caused by this congestion. These ex-

ternalities are not only in the form of environmental

impacts but are also of an economic nature: A consid-

erable proportion of federal expenditures for improving

the air traffic control system goes to improving the traffic-

handling capacity in and around the relatively few major

hubs where most of the congestion occurs. Given that

dehubbing is not practical (nor necessarily desirable), it

would stand to reason that a hubbed system might be

rationalized by some form of pricing mechanism.

With this strong tendency toward hubbing, it would

seem that airports might benefit from being selected as

hubs by airlines, especially if the concomitant revenue

potential were exploited. Recent trends in the post de-

regulation era show increased hubbing of airline net-

works. This suggests two things: One is that some air-

ports are losing traffic; and the other is that a smaller

number of hub airports are gaining this traffic. If this

trend continues, then one would expect that some of the

I

i

AIRCRAFT FLYING COSTS

Z

i

[I

Z

HUB PENALTY

I

,

360

720 1080

1440 181

PENALTY LEVEL

Fig. I I. Hub penalty and airline operating cost.



27

irline hubbing and airline economics

360

720 1080

1440

18(

PENALTY LEVEL

Fig. 12. Hub penalty and airport revenue.

hubs would become congested to the point where they

will begin to lose their attractiveness, and as a conse-

quence new hubs might emerge. This can already be

observed in the system at such places as Kansas City

and Memphis. There are many such airports in the sys-

tem, and it is a fair bet that the majority of them would

welcome the opportunity to become an airline hub. This

suggests that airports of currently medium size will be

competing for airline service and will have to begin to

“market” themselves as convenient transfer facilities.

What this competitive behavior might lead to in terms

of network structure, and how it will affect airport eco-

nomics, are interesting subjects for further research. But

it is clear at present that airline hubbing is not necessarily

inefficient and that it is likely to persist as an important

feature for air transportation systems.

REFERENCES

Daganzo C. F. and Spamann J. (1982)

TROMP User’s Manual:

Calibration Predicrion and Confidence Intervals.

Inst. Transpn.

Studies, Berkeley, CA UCB-ITS-RR-82-4, 89 pp.

deNeufville R. E. and Gordon S. (1972) Design of air trans-

portation networks.

Trunspn Res. 7: 3 207-222).

Ghobrial A. (1983) Analysis of the air network structure: The

hubbing phenomenon. Dissertation, Univ. California, Berke-

ley, UCB-ITS-DS-83-2, 138 pp.

Gosling G. D. (1979) An economic framework for the planning

of airport passenger terminals. Dissertation, Univ. California,

Berkeley, UCB-ITS-DS-79-1, 401 pp.

Kanafani A. and Ghobrial A. (1982) Aircraft evaluation in air

network planning.

Tronspn. Eng. J.

ASCE, 108: TE3 (282-

300).

Kanafani A. (198 I) Aircraft technology evaluation and network

structure

in short haul air transportation. Transpn. Res. 15A:

4 (305-314).

Airline hubbing—Some implications for airport economics.pdf

Documents