entry deterrence and quality provision in the local bus market

This article was downloaded by: [The University of Manchester Library]On: 16 October 2014, At: 03:03Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,UK

Transport Reviews: ATransnational TransdisciplinaryJournalPublication details, including instructions for authorsand subscription information:http://www.tandfonline.com/loi/ttrv20

Entry deterrence and qualityprovision in the local bus marketJan Peter Van Der VeerPublished online: 26 Nov 2010.

To cite this article: Jan Peter Van Der Veer (2002) Entry deterrence and quality provisionin the local bus market, Transport Reviews: A Transnational Transdisciplinary Journal,22:3, 247-265, DOI: 10.1080/01441640110115092

To link to this article: http://dx.doi.org/10.1080/01441640110115092

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all theinformation (the “Content”) contained in the publications on our platform.However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, orsuitability for any purpose of the Content. Any opinions and views expressedin this publication are the opinions and views of the authors, and are not theviews of or endorsed by Taylor & Francis. The accuracy of the Content shouldnot be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions,claims, proceedings, demands, costs, expenses, damages, and other liabilitieswhatsoever or howsoever caused arising directly or indirectly in connectionwith, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes.Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly

http://www.tandfonline.com/loi/ttrv20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/01441640110115092

http://dx.doi.org/10.1080/01441640110115092

forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Entry deterrence and quality provision in the local bus market

JAN PETER VAN DER VEER*

National Economic Research Associates, 15 Stratford Place, London W1C 1BE, UK

(Received 9 May 2001, revised 11 October 2001; accepted 11 November 2001)

In a deregulated bus market, incumbent operators often seek to deter entry bysetting frequency levels to avoid leaving pro®table gaps. The consequences of thisaction have been analysed using a simulation model of a hypothetical incumbentbus operator vulnerable to entry. The model features two dimensions of quality: a`horizontal’ frequency dimension and a `vertical’ quality dimension. It isexamined whether, from a social point of view, such entry deterrence strategieslead to oversupply in the horizontal frequency dimension; what the consequencesof this are for the supply of vertical aspects of quality; and what the impact ofregulation and quality/output related subsidies would be. The results suggest thatan incumbent operator will indeed oversupply in the frequency dimension to deterentry. It undersupplies in the quality dimension, though, but supplies morequality than would have been done in the absence of an entry threat. Qualityregulation and quality subsidies can lead to modest welfare gains, but nowherenear those that can be achieved using price regulation or output subsidies. Ifpaying subsidies results in further oversupply of bus miles, the welfare eVects ofthe subsidies may be limited or even negative.

1. IntroductionIn the 10 Year Plan `Transport 2010’, the UK government has set itself a target

of achieving a 10% growth in bus passenger journeys over the next 10 years, almost1% a year. The government regards public transport as vital in tackling pollutionand congestion, in giving people more choice in how they travel and in achieving afairer society in which everyone has reasonable access to transport. Every year, some4.3 billion passenger journeys are made by bus, representing over two-thirds of allpublic transport journeys. The bus is, therefore, an important element of the UKtransport system. During the past 50 years, however, its use has been decliningsteadily at about 2.5% per annum. Consequently, the number of local bus passengerjourneys is now only about 25% of what is was in 1950.

The 10 Year Plan contains detailed proposals on how to reverse this decline andto achieve the government’s target. It builds on the ideas published in FromWorkhorse to Thoroughbred: A Better Role for Bus Travel (1999). Some of themeasures are external to the bus industry but will have a major impact on it, likeroad-user charging and workplace parking levies. Other proposals relate directly tothe bus market, such as better information, frequency enhancements, more localauthority in¯uence over buses and minimum half-fare concessions for the elderly.

*e-mail: [email protected] m

Transport Reviews ISSN 0144-1647 print/ISSN 1464-5327 online # 2002 Taylor & Francis Ltdhttp://www.tandf.co.uk/journals

DOI: 10.1080/01441640110115092

TRANSPORT REVIEWS, 2002, VOL. 22, NO. 3, 247±265

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

Improving the quality of bus services is seen as vital and the government believesthat Quality Partnerships are part of the solution.

Quality Partnerships are agreements between one or more operators and a localauthority in which the partners jointly undertake measures to improve the quality ofbus services. Typically, this involves investment on behalf of both parties. The localauthority might invest in bus priority schemes, new shelters or more radical solutionslike bus lanes or guided busways. The operator would normally provide new buseswith easy access and environmentally friendly characteristics. Over 30 towns andcities have already adopted the approach.

There are, however, some potential problems. One of these is the free-riderproblem: how to prevent operators who do not participate in the agreementfrom using any new infrastructure provided? If quality operators could beundercut by low-cost, low-quality operators, they might be reluctant to makeany investment at all. The Transport Act 2000, therefore, provides a statutorybacking for quality partnerships, which will allow local authorities to setquality standards for facilities they provide and to exclude operators who donot meet these standards from using the facilities. For areas where QualityPartnerships might not work, the Act also includes the option of adopting thecurrent London model of franchising bus services in the form of QualityContracts. Quality Contracts are, however, subject to ministerial consent andthe government initially expects to grant only a limited number of approvals toallow the approach to be evaluated.

The emphasis on quality in the government’s policy appears to stem from a beliefthat the current bus operators fail to provide the desired quality levels. However, it isoften thought that one of the most important dimensions of quality, frequency,actually tends to be oversupplied in a free bus market as a result of the nature ofcompetition (e.g. Evans 1987). An incumbent monopoly operator, for example,might oversupply in the frequency dimension to avoid leaving gaps in whichcompetitors could enter. It is di� cult to predict what the consequences of this wouldbe for the supply of `other’ dimensions of quality.

The present paper will therefore seek to answer three main questions.

. To what extent will an incumbent monopoly operator oversupply in thefrequency dimension to deter entry?

. What are the consequences of the entry deterrence strategy for the supply of`other’ quality dimensions?

. Can regulation be used to achieve a more favourable outcome?

The remainder of the paper is set out as follows. Section 2 discusses the conceptof quality in the local bus market and argues why it is necessary to distinguishbetween multiple dimensions of quality. Section 3 presents a simulation model of anurban bus operator. Section 4 deals with the results and contains the basic outcomes,the impact of entry deterrence strategies for the supply of quality and the potential ofregulation. The paper concludes with a discussion in Section 5.

2. Quality in the local bus marketThe concept of quality of bus services is complex as there are several factors and

actors that determine the level of quality a customer will perceive. It is worthdistinguishing between the following determinants of perceived service quality.

248 J. P. van der Veer

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

. The quality of the operator the customer is travelling with. This aspect will bediscussed in more detail below.

. The quality of any competing operators. There may be some kind of externalitybetween operators if customers regard the quality of operator A to be diVerent ifthere are buses of a diVerent quality competing on the same route.

. The quality of the services provided by the local authority. This usually includesshelters and timetable information, but also the tra� c environment andinfrastructure priorities. DiVerent local authority departments may be respon-sible for diVerent aspects, e.g. the transport executive as opposed to the highwayagency.

. Other customers travelling on the same bus. Boarding times will be longer andthe journey will be less pleasant if the bus is fuller. This is a two-way process as acustomer also causes quality for the other customers to decline.

. Other road users. The busier the road, the more likely the bus is to be stuck intra� c. In the short-term, this is a two-way process only to the extent that longerboarding times because of an additional customer cause queues for other tra� c.In the longer-term, more bus passengers will also cause the number of buses toincrease and then the externality clearly works both ways.

. The expectations of the customer. Not only will expectations in¯uenceperceptions, but also perceived service quality will in¯uence expectations offuture uses of the service. This is therefore also a two-way process.

The determinants are shown in ®gure 1.The paper focuses on the quality of service from the perspective of the operator

the customer is travelling with. Even with this simpli®cation, however, it is stillnecessary to distinguish between two fundamentally diVerent dimensions of quality.

If a bus operator diVerentiates its service from the one oVered by anotheroperator by using more comfortable vehicles, providing better information andhiring better quali®ed staV, all customers will prefer this service to the other, all otherthings being equal. These dimensions of quality enable an operator to diVerentiate itsproduct vertically (Abbott 1953). By contrast, if an operator diVerentiates its serviceby oVering diVerent departure times or diVerent routes than its competitor,customers will not make a unanimous choice, as people live at diVerent places andwish to depart or arrive at diVerent times. Hence, diVerentiation between operatorsaccording to routes or departure times oVered is an example of horizontal productdiVerentiation.

The reason that these dimensions are fundamentally diVerent is the waycompetition in the bus market works. When determining the optimal level of vehiclecomfort, information provision, etc. to oVer, an operator will look at the marginalcosts and revenues of the quality aspect under consideration. By contrast, in aderegulated bus market, the marginal costs and revenues of changes in departuretimes or routes are not the only aspect to consider. An incumbent operator will alsoconsider whether a timetable or a route schedule leaves any entry opportunities for apotential competitor. Given the ability of incumbent bus operators to change pricesinstantaneously when entry occurs, it is not normally a feasible strategy for anentrant to gain market share by charging lower fares. The only way to compete is inthe frequency dimension. If an operator’s pro®t-maximizing schedule would involvegaps in which a competitor could pro®tably enter, the incumbent may well furtherexpand its services to avoid leaving such gaps. The short-term pro®t consequences of

Entry deterrence and quality provision in the local bus market 249

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

Figure 1. Determinants of perceived service quality in the local bus market.


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

such an expansion will then be of secondary importance. Consequently, equilibriumin a deregulated bus market is likely to involve higher fares and higher frequencylevels than socially optimal (Evans 1987).

Schmalensee (1978) investigates the issue in the context of brand proliferationin the breakfast cereal market to deter entry. Of particular relevance is hisanalysis of the welfare properties of alternative equilibria. Though Schmalenseeshows that the number of brands in the deterrence equilibrium exceeds thenumber of brands in the monopoly position, he also shows that given monopolypricing, it is actually optimal to increase the number of brands above themonopoly position. Crucially, it cannot be determined in advance whether thenumber of brands in the deterrence equilibrium is less or more than the welfare-maximizing position, the welfare-maximizing number of brands given monopolyprices or the welfare-maximizing number of brands given a zero-pro®tconstraint. It should be noted that the deterrence equilibrium involves positivepro®ts as a result of the imperfect competitiveness of the industry and thatprices will be higher than optimal. Furthermore, he shows that if pro®ts aredriven to zero, it is best to do so with a price below the monopoly level. Inaddition, he proves that the best zero-pro®t point is strictly better than thedeterrence equilibrium. The implication of this is that the problem with thedeterrence equilibrium is not that too many brands are introduced but ratherthat too little price competition occurs. However, it is only possible to provethat some increase in price competition will improve welfare, not that allincreases in price competition would do so. The problems in terms of limitedinformation available to the regulators are obvious.

The bus industry cannot be regarded as perfectly contestable (Mackie andPreston 1996). However, even if it were, the deterrence equilibrium in the busindustry would involve positive pro®ts. The reason for this is that headways cannotbe adapted continuously as a result of indivisibilities in the number of buses per timecycle. Even at competitive price levels, therefore, superpro®ts will be made which areby no means negligible, although nowhere near the monopoly level (Evans 1987).The entry barriers that exist in the local bus market (Mackie and Preston 1996) couldlead to even larger pro®ts being made as an incumbent operator does not need toexpand services as much as it would have done in the absence of entry barriers. Someoverprovision of services, compared with the monopoly level, is still likely and pro®tswill therefore still be lower than what a fully protected monopolist would achieve.Following Schmalensee’s analysis discussed above, it cannot be determined inadvance whether the deterrence equilibrium would involve fewer or more servicesthan socially optimal, although prices will be higher. In the bus industry, too, theproblem might rather be a lack of price competition than an overprovision ofservices.

3. Simulation of an urban bus routeThis section develops a simple simulation model of a hypothetical bus route that

incorporates both a `horizontal’ and a `vertical’ dimension of quality. The terms`frequency’ or `frequency dimension’ will refer to quality in the sense of horizontalproduct diVerentiation, whereas the terms `quality’ or `quality dimension’ refer to allother quality aspects that are vertically diVerentiated. The model will be used toassess to what extent an incumbent operator will oversupply in the frequencydimension to deter entry; the impacts of this for the supply of the other aspects of


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

quality; and whether regulation or subsidy systems can be used to achieve a morefavourable outcome.

The model features an incumbent ®rm operating a hypothetical bus routethrough a major city. The route is treated as self-contained with eight zones andseven stages between the zones. There is only one ticket type, which is oVered at a ¯atfare. The operator is vulnerable to entry and will therefore set frequency levels toavoid leaving pro®table gaps. It will set fares at pro®t-maximizing levels.

The model consists of three submodels: a demand model, a cost model and asocial welfare model. These are discussed in turn.

The demand model features two quality variables: one relating to frequency andthe other relating to all other aspects of quality such as comfort, vehicle accessibilityand staV friendliness. As it is di� cult to quantify some of these aspects and as theevidence on the valuation of quality measures in the bus industry is scarce, the latterquality variable has been de®ned as the valuation of the quality aspects. (Essentially,this assumes that the operator determines the optimal valuation of quality thatshould be achieved and then oVers the physical quality level necessary to achievesuch a valuation.) The third variable in the model is a price variable. Following somesimulation, it emerged that it would be appropriate to include two interactionvariables: between price and buses, and between buses and quality. The simulationsuggested that a negative exponential model form would best match the availabledata and hence the following form has been used:

Q ˆ y¤exp…aP ‡ c

B‡ dZ ‡ eBP ‡ fBZ†; …1†

whereQ number of passengers (thousands/year),P ¯at fare (£),B number of buses per h,Z valuation of the level of quality oVered (£), anda, c, d, e, f, y parameters to be estimated.Using the available simulation data, the model has been calibrated to give an

overall elasticity of demand of ± 0.4.The cost model has been built using the CIPFA formula, allocating various types

of costs to vehicle-hours, vehicle-km or peak vehicles (CIPFA 1974). FollowingWhelan et al. (2000), it was assumed that 75% of total costs are related to vehicle-hours, 10% to vehicle-km and 15% to peak vehicles. However, applying the CIPFAformula would result in a cost function that would not vary with the number ofpassengers carried, in other words the marginal cost with respect to passengernumbers would be zero. To avoid this unrealistic property in the model, a distinctionwas made between vehicle-hours due to the boarding of passengers, which do varywith the number of passengers carried, and all other vehicle-hours. The basic costmodel is therefore as follows:

TOC ˆ gVh ‡ iBh ‡ jVkm ‡ kVreq; …2†

whereTOC total operating cost per day,Vh vehicle-hours excluding boarding time,Bh boarding-hours,Vkm vehicle-km,


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

Vreq peak vehicle requirement, andg, i, j, k parameters to be estimated.Although there is now a non-zero marginal cost, as more passengers increase the

load factor that in¯uences the boarding and alighting time penalty, the model stillassumes no capacity restrictions. In other words, the marginal cost for the secondpassenger is the same as for the 52nd passenger, which is clearly not realistic. Strictlyspeaking, this is true both for marginal operator cost as well as marginal user cost,because passengers also impose costs on each other. However, incorporation ofincreased user costs because of more passengers would require a complicatedadaptation of the demand function (1).

Instead, a method has been chosen where the operator is forced to run reliefbuses as soon as the load factor exceeds a certain level. EVectively, this assumes thatall additional user costs are absorbed by the operator, who simply prevents busesfrom becoming full or overcrowded. Using data on the spread of passengers acrossthe day on bus lines, an estimate has been made of the number of buses likely to beovercrowded, given some average load factor. The estimates have been used toestimate a quadratic function of relief costs, with the average load factor asexplanatory variable.

Finally, the costs of providing quality measures have been added to the model,initially solely as operator costs. Not only is the evidence on the valuation of qualityin bus services scarce, but also it is even more di� cult to link quality valuations toactual amounts of money spent on quality. It might be reasonable to assume,though, that small bene®ts can be achieved with relatively simple measures likeinformation provision and that progressively more expensive projects are needed toachieve higher values. Quality costs have therefore incorporated in the model as aquadratic function of quality Z. They have been assumed ®xed, spread over thelifetime of the facility by means of some depreciation method.

The cost model has been calibrated assuming an average operating cost of £0.90per km, an average load of 10 (Whelan et al. 2000), and a boarding and alightingtime penalty of 3 s per passenger (Cundill and Watts 1973). The fact that the averageoperating cost is given implies that vehicle size in the model is ®xed. We return to thisissue in the discussion.

Pro®t is de®ned as:

P ˆ PQ ¡ TC; …3†

whereas consumer surplus can be derived from the demand function (1) as:

CS ˆ Q…a ¡ e¤B† : …4†

A shadow cost of public funds of 1.3 has been assumed. Therefore, assuming thatthe operator will be subsidised if it is unpro®table, welfare, W, is de®ned as:

W ˆ CS ‡ PQ ¡ TC if …PQ ¡ TC†50

W ˆ CS ‡ 1:3…PQ ¡ TC† if …PQ ¡ TC†50…5†

4. ResultsThe model developed above can be solved by varying the input variables B (buses

per hour), P (¯at fare, £) and Z (level of quality) to ®nd maximum attainable values


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

for such variables as pro®t, welfare and the number of passengers. The inputvariables are subject to some constraints, the most important of which being therequirement that the number of buses per hour must be a positive integer (see Section2). Quality and price are constrained to be non-negative. Other constraints can beadded as desired, such as a break-even constraint.

Some basic results, including the pro®t and welfare-maximizing positions, arepresented in table 1. The format of table 1 is the general format in which results willbe presented and includes:

P ¯at fare (£),B number of buses per hour,Z quality; value of the operator speci®c coe� cient (£),Q number of passengers (thousands per year),P pro®t (thousands of £ per year),W welfare (thousands of £ per year, including any subsidy cost),S subsidy (thousands of pounds per year) andep price elasticity.A full monopoly (i.e. protected by statutory entry barriers) will undersupply both

in the frequency and quality dimensions. The pro®t-maximizing price is more thanfour times as high as the welfare-maximizing price, with a resulting deadweight lossof almost 50%. The welfare-maximizing price is much higher than the marginal cost,which is around £0.04, depending on the average load factor. It is the price where,for marginal changes, the changes in consumer surplus are equal to the change inoperator losses, multiplied by the shadow cost of public funds. It is highly sensitiveto the assumed shadow cost: assuming it to be 1.4 instead of 1.3, for example, wouldresult in an optimal price of £0.26 instead of £0.21.

When a breakeven constraint is introduced, the optimal price more than doubles,but the welfare loss is relatively limited. Maximizing passengers subject to thisconstraint, as opposed to maximizing welfare, implies a lower price and fewer buses,but the same quality level. Maximizing quality subject to a breakeven constraintinvolves charging monopoly prices to maximize revenue and then spending this onquality.

4.1. Deterrence equilibriumIn Section 2, the likely properties of the deterrence equilibrium in the bus

industry were discussed. It was shown there that the deterrence position wouldinvolve positive pro®ts as a result of the ®nite number of buses per time cycle. It

Table 1. Basic results.

Objective P B Z Q P W S ep

Maximum pro®t 0.85 3 0.038 1220 375 1387 0 71.03Maximum welfare 0.21 8 0.293 5211 71583 2532 0 70.24Maximum welfaresubject to breakeven

0.47 5 0.118 2639 0 2241 0 70.56

Maximum passengerssubject to breakeven

0.39 4 0.118 2666 0 2236 0 70.47

Maximum qualitysubject to breakeven

0.85 4 0.241 1804 0 1514 0 71.02


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

cannot be determined in advance whether the deterrence equilibrium results inunder- or overprovision of services.

Table 2 shows the pro®t-maximizing positions when the number of buses isexogenously given. It is assumed here that the market determines the level of qualityand that all quality costs are borne by the operator. From table 2, it can be seen thatwhen prices are set at pro®t-maximizing levels, it is not possible to run a pro®tableservice with seven (or more) buses per hour. Should prices be lower, e.g. if two ormore operators were operating on the route, the maximum frequency that can beoperated pro®tably would be lower too.

The best entry-deterring strategy for the incumbent operator would thereforeappear to be operating a service with six buses an hour. Any additional service wouldbe unpro®table. The deterrence position does indeed involve pro®ts that are notnegligible, as suggested by Evans (1987).

In practice, entry barriers may exist as a result of asymmetries between theincumbent and entrant. An ability of an incumbent operator to introduce apro®table sixth service does not imply that an entrant will do so as well, e.g. becauseof diVerences in access to facilities or consumer awareness. Consequently, theincumbent operator could limit its service to just ®ve buses per hour, for example,thereby earning even larger superpro®ts. We will, however, assume that there are nosuch asymmetries and that the monopolist deters entry by running six buses perhour. Table 1 shows that by doing so, it clearly runs more buses than a fullyprotected monopolist would have done. It also oversupplies buses compared with thebreak-even welfare-maximizing outcome, but not compared with the full welfare-maximizing position.

An interesting property of the deterrence equilibrium is the fact that the price ishardly diVerent from the full monopoly price. This can be understood by looking atthe expression for the pro®t-maximizing price, which can be derived from thedemand and pro®t functions (1) and (3):

P ˆ 1

a ¡ e*B‡ MC: …6†

Given an a of around 1.25 and e of around 0.014, it can be seen that the impact ofB is rather small. Furthermore, it can be shown that it is partially oVset by the natureof the marginal cost function. The monopolist will therefore charge virtually thesame price, whatever the number of buses oVered.

Table 2. Maximizing pro®ts subject to B = n, n = 4 ± 7.


Maximum pro®tsubject to B = 4

0.85 4 0.051 1415 325 1511 0 71.02


0.86 5 0.061 1553 231 1550 0 71.01


0.87 6 0.070 1659 113 1538 0 71.01


0.87 7 0.078 1744 719 1492 0 71.01


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

An important issue is obviously the level of quality the monopolist chooses toprovide. The ®rm will ®nd it optimal to supply almost twice as much quality as itwould otherwise have done. However, it will still undersupply in the qualitydimension, both compared with the welfare-maximizing position and with the break-even welfare-maximizing outcome.

The entry-deterring strategy results in a welfare improvement of around 11%compared with the full monopoly position. However, it comes nowhere near thewelfare levels that would be attainable given the number of buses run, even if abreak-even constraint is imposed (table 3).

Given the large gap between the welfare levels reached in the deterrenceequilibrium and the potentially attainable welfare levels, the question arises whetherregulation, subsidy systems or some combination of both could bring bene®ts. Wewill explore some possibilities below.

4.2. Quality regulation without paying subsidyRegulating quality would involve setting some quality level with a statutory

requirement for the operator to comply. It would be an extension of, for example,existing safety requirements. An important question for the regulator is of coursewhat level of quality to set.

The above analysis showed that the entry-deterring monopolist chooses toprovide more quality than a fully protected monopoly would do, but less than theoptimal level of quality of the optimal level given a break-even constraint. A ®rstthought might be to set the standard at the latter level and see how the operatorwould respond. As running six buses per hour is still pro®table at this level ofquality, this will remain the deterrence equilibrium. Table 4 shows the results. Itshould be noted that the number of buses when pro®ts are maximized in this and allfollowing tables is the number of buses the operator would choose to operate as todeter entry while still earning a positive pro®t or breaking even.

Although table 4 shows that welfare has improved as a result of the qualitystandard, the resulting levels are by no means those achieved in table 1, for example.The reason for this is the fact that both price and the number of buses remainine� ciently high. The optimal quality standard therefore needs to be determined

Table 3. Welfare levels attainable given B = 6.


Maximum welfaresubject to B = 6

0.21 6 0.227 4391 7973 2509 0 70.24

Maximum welfare subject 0.59 6 0.114 2421 0 2081 0 70.69to B = 6, breakeven

Table 4. Regulating quality.


Maximum pro®tsubject to Z = 0.118(deterring entry)

0.87 6 0.118 1768 93 1612 0 71.01


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

given a service with six buses per hour and monopoly pricing. Table 5 gives theresults.

Table 5 shows that the optimal quality standard under overprovision of buses ismuch higher than the quality level of the second-best breakeven position. The higherquality level results in a modest welfare improvement over the position in table 4, butwelfare is still nowhere near the levels shown in table 3 to be attainable, even under abreakeven constraint.

A relevant question is what would happen if the regulator gets the qualitystandard wrong. Any standard from 0.070 (the level the entry-deterring ®rmwould choose) up to 0.184 will increase welfare compared with the deterrenceequilibrium level. Table 6 shows what would happen if quality levels are setabove the optimum.

For quality levels 40.184, operating six buses per hour is no longer pro®tableand the deterrence equilibrium will shift to ®ve buses per hour. The welfare losses ofthe inappropriate quality standard are still limited, though, as they are almost oVsetby the elimination of oversupply in the frequency dimension. Although the situationis not optimal, it is still better than the deterrence equilibrium. However, when thequality standard would be set 40.223, the operator will only run four buses per hourand there will then be welfare losses compared with the situation without any qualityregulation at all.

As already mentioned, the quality standard has to be set high because both pricesand frequency levels are at ine� cient levels. However, if the deterrence equilibriumwould involve running less than six buses per hour as a result of entry barriers, theoptimal standard would be actually be even higher. The reason for this is the factthat running fewer buses involves greater pro®ts. Social welfare can to a certainextent be increased by using these pro®ts to achieve a higher quality level.

Table 5. Optimal quality regulation.


Maximum welfare subjectto P = 0.87, B = 6

0.87 6 0.184 1933 0 1661 0 71.01


0.87 6 0.184 1931 0 1659 0 71.01

Maximum welfaresubject to P = 0.87

0.87 6 0.184 1933 0 1661 0 71.01

Table 6. Setting quality standards too high.



0.86 5 0.190 1847 85 1653 0 71.01

Maximum pro®t subjectto Z = 0.23(deterring entry)

0.86 4 0.230 1777 38 1528 0 71.02


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

4.3. Price regulation without paying subsidyIt was shown above that the monopolist charges virtually the same price, around

£0.87, whatever the level of frequency and quality oVered. Table 5 showed that themaximum attainable welfare given this price level is rather limited and priceregulation would therefore appear to be an obvious option.

An operator that deters entry and faces a regulated price has eVectively only thelevel of quality at its discretion. For each level of the regulated price, it chooses themaximum number of buses that yields a positive pro®t and subsequently maximizesthe level of pro®t by adopting a quality level. Table 7 contains the pro®t-maximizingdecisions for a range of regulated price levels.

Though tougher price targets cause the operator to economize both on qualityand on frequency levels, the welfare improvements are very substantial . The optimalfare is just above £0.43, where the operator will break even running the second-bestoptimal service of ®ve buses per hour. For lower fares, the operator will have to cutfrequency levels further and welfare will drop. Compared with quality regulation,however, any fare between about £0.35 and £0.60 will achieve superior results, andthe consequences of a wrong standard within this range are only minor.

4.4. Price and quality regulation without paying subsidyA remaining option is to regulate price and quality at the same time to avoid the

cuts in quality that were seen above to be the result of price regulation. In this case,the operator would not do anything else than setting frequency levels as to deterentry. The regulatory authority will now attain the second-best solution subject to a

Table 7. Regulating price.


Maximum pro®tsubject to P = 0.80(deterring entry)

0.80 6 0.070 1791 109 1648 0 70.93


0.56 6 0.062 2329 2 2004 0 70.66


0.50 5 0.050 2335 74 2057 0 70.59


0.43 5 0.044 2506 0 2128 0 70.51


0.40 4 0.033 2376 75 2067 0 70.48


0.35 4 0.028 2506 4 2106 0 70.42


0.30 3 0.014 2304 23 1933 0 70.36


0.28 2 0.000 1814 37 1524 0 70.34


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

breakeven constraint from table 1. This involves a fare of £0.47 and a quality level of0.118. The only way in which the operator can deter entry given these constraints isto run the optimal service of ®ve buses per hour. The welfare improvement comparedwith the deterrence equilibrium without any regulation at all is 445%.

4.5. Paying lump-sum subsidiesA crucial question in the context of paying lump-sum subsidies to an entry-

deterring incumbent operator is whether the subsidy will aVect the deterrenceequilibrium. This depends on whether the operator expects the subsidy to beavailable, in some proportion, to new entrants as well. It will be assumed that thiswill be the case and that, therefore, the incumbent operator will base its deterrencestrategy on the level of pro®ts including any subsidy received. This means that if theoperator currently runs six buses per hour at £40 000 annual pro®t, whereas sevenbuses per hour would lose £160 000 per year, a £200 000 lump-sum subsidy wouldmake seven buses per hour `pro®table’ and this would therefore be the newdeterrence equilibrium.

Paying a lump-sum subsidy of, for example, £200 000 per year without anyfurther regulation will result in substantial welfare losses (table 8). The deterrenceequilibrium will now involve even more oversupply in the frequency dimension, eightbuses per hour, with only a slightly higher quality level. Not just this but all levels oflump-sum subsidy will result in welfare losses, as a result of the shadow cost ofsubsidy (still assumed to be 1.3) and, potentially, increased oversupply of buses.

The optimal quality standard given eight buses per hour and this subsidy level is0.154. This level happens to be lower than at the unsubsidized deterrence equilibriumas this particular subsidised deterrence equilibrium involves lower pro®ts (38 asopposed to 113), and there is therefore less money available to invest in quality.(When the lump-sum subsidy would be £300 000 as opposed to £200 000, forexample, the optimal quality standard would be 0.215, which is higher than in theunsubsidized deterrence equilibrium.) However, as this frequency level is ine� cientlyhigh, it is actually better to set a much higher quality standard. This will reducepro®ts and the operator will respond by reducing the oversupply in the frequencydimension. The optimal quality standard given a £200 000 lump-sum subsidy is0.261, which is the level at which the subsidised operator just breaks even running sixbuses per hour. This frequency level is optimal under monopoly pricing (table 5).However, compared with table 5, welfare is considerably lower as a result of

Table 8. Paying an annual £200 000 lump-sum subsidy.


Maximum pro®t(deterring entry)

0.88 8 0.086 1816 38 1378 200 71.00


0.88 8 0.154 1972 1477 200 71.01


0.87 6 0.261 2137 1576 200 71.01

Maximum pro®t P = 0.31(deterring entry)

0.31 5 0.030 2863 2171 200 70.36

Maximum welfare subjectto breakeven

0.36 5 0.138 3091 2365 200 70.42


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

excessive quality and the shadow subsidy cost. Consequently, even in combinationwith quality regulation, lump-sum subsidies result in welfare losses.

The picture may change when looking at the impacts of price regulation incombination with a lump-sum subsidy. For any regulated price down to £0.43, theoptimal price standard in the absence of subsidy payments, the lump-sum subsidystill results in welfare losses as a result of overprovision in the frequency dimensionand the shadow subsidy cost. For price standards below £0.43, where theunsubsidized operator would only run four buses per hour, the lump-sum subsidyresults in welfare gains. The optimal price, as in the unsubsidized situation, is the fareat which the operator just breaks even running the second-best optimal ®ve buses perhour. This fare is £0.31 and the result is also shown in table 8.

Finally, when both price and quality are regulated, the authority will attain thesecond-best solution given a £200 000 annual loss. The optimal price then becomes£0.36 with a quality standard of 0.138. This combination yields a substantial welfarelevel.

It should be noted that a modest lump-sum tax would result in welfare gains,both because pro®ts can be transferred into public funds (with a shadow revenue of1.3) and because an appropriate tax can shift the deterrence equilibrium to a moree� cient ®ve buses per hour.

4.6. Paying quality-related subsidiesGiven the often negative eVects of lump-sum subsidies as described above, the

authority may instead choose to link subsidies to some measure of performance. Thissubsection will examine what happens if subsidies are linked to the level of qualityoVered. It will be assumed that the authority subsidizes the operator according to theformula k*Z. Table 9 shows the results with k in £ thousands of pounds.

Table 9 shows that the eVects of paying a quality subsidy are ambiguous. Thougha modest subsidy does lead to small welfare gains, the point is soon reached where

Table 9. Paying a quality-related subsidy.


Maximum pro®tsubject to k = 0(deterring entry)

0.87 6 0.070 1659 113 1538 0 71.01


0.87 6 0.082 1684 128 1554 16 71.01


0.87 7 0.096 1786 7 1523 29 71.01


0.87 7 0.149 1919 116 1554 178 71.01


0.88 8 0.164 2022 1 1506 213 71.00


0.88 8 0.206 2145 130 1484 412 71.00


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

the entry-deterring operator expands from six to seven buses an hour. At this point,the bene®ts of the increase in quality are more than oVset by the costs of increasedoversupply of buses. Increasing the quality subsidy even further will again result inwelfare gains until the number of buses becomes eight. From there on, qualitysubsidies result in both excessive quality and frequency levels and consequently inwelfare losses.

4.7. Output-related subsidiesAn alternative way of paying subsidies is to link them to a measure of output,

which could be either bus- or passenger-miles. Subsidizing bus miles in a market thatalready tends to oversupply them does not appear to be very sensible and we willtherefore focus on paying a subsidy on the basis of the number of passengers carried.

An important feature of an output-related subsidy is that it causes a pro®t-maximizing monopoly to set prices below the monopoly level. It can be shown that ifthe subsidy paid is an amount k per passenger and the demand function is of theform Q = exp( ± bP), price will be set at the monopoly level minus k. (In the case of alinear demand function Q = a ± bP, the diVerence between the monopoly level andthe price the operator will set is ± 0.5k.) However, pro®ts increase as a result of thesubsidy and the deterrence equilibrium may therefore change.

Table 10 shows the eVects of paying various levels of output subsidy. The levelsshown are the lowest and highest levels of subsidy per passenger which are consistentwith a given deterrence equilibrium.

It can be seen that rising subsidy levels (per passenger) imply lower prices, andalso higher quality and, generally, higher welfare. However, the eVects of the pricesand quality are largely oVset by increased oversupply in the frequency dimension.The bene®ts per pound of subsidy are therefore not very large.

4.8. Subsidy systems in combination with lump-sum taxesAll subsidy systems examined so far tend to cause even more oversupply in the

frequency dimension than was already the case in the original deterrenceequilibrium. This oVsets the welfare gains or even leads to net welfare losses. Wewill look in this subsection at ways of preventing this by combining subsidy systemswith lump-sum taxes. As already mentioned, lump-sum taxation can result in welfaregains even without subsidy systems.

It will be clear that by setting the appropriate combinations of output and qualitysubsidy and removing any pro®ts by lump-sum taxation, the authority will achievethe second-best outcome, or even more if the authority is prepared to pay a netsubsidy. This, however, poses great information requirements for the regulator as theoptimal number of buses may diVer as price and quality are varied. Given highprices, it is actually optimal to run six buses per hour. This has been shown in table 5for the monopoly level and is the case for all prices down to £0.64. When a break-even constraint is imposed, the optimal number of buses for a wide range of faresbelow £0.64, down to £0.44, becomes ®ve. When no break-even constraint isimposed, the optimal number becomes higher, up to eight, as the ®rst-best solution isreached, though at the expense of huge subsidies. When the authority is prepared topay a net subsidy of £200 000 per year, for example, the optimal number of busesremains six down to a fare level of £0.45 and becomes ®ve afterwards.

When subsidizing quality only, prices will remain at the monopoly level and theoptimal number of buses given this price level is six. This can be reached by paying


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

an appropriate quality subsidy and removing the pro®ts by lump-sum taxation. Theresult will be equivalent to directly regulating quality, with the welfare improvementbeing limited by the continued high fares.

Using output-related subsidies is more powerful. The authority can reach theoptimal situation given six buses an hour and a break-even constraint by paying anappropriate output subsidy and removing the resulting pro®ts. However, at theresulting fare of £0.59, the optimal number of buses per hour is actually ®ve and thesituation is not therefore optimal. A better option is therefore to oVer a higheroutput subsidy, bringing the price to the second-best optimal level of £0.47, and notjust to remove the resulting pro®ts but to impose an even higher tax, making sixbuses per hour unpro®table and shifting the deterrence equilibrium to ®ve buses anhour.

It is worth noting that where quality-related subsidies result in higher quality andproduce an equivalent result to quality regulation, output-related subsidies result inlower prices but are better than price regulation. The reason for this is the fact that ata given price, the total revenue per passenger in combination with the output subsidy

Table 10. Paying an output-related subsidy.


Maximum pro®tsubject to outputsubsidy = 0.02(deterring entry)

0.85 7 0.081 1791 16 1528 36 70.98


0.80 7 0.090 1942 128 1615 155 70.91


0.79 8 0.100 2052 12 1580 185 70.90


0.74 8 0.109 2199 119 1655 308 70.85


0.74 9 0.121 2311 3 1614 347 70.83


0.69 9 0.132 2481 123 1691 496 70.78


0.69 10 0.144 2602 11 1651 546 70.77


0.65 10 0.155 2761 118 1715 690 70.72


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

will still be equal to the monopoly level. The output subsidy also applies to anyincrease in passengers as a result of a quality improvement. Price regulation, bycontrast, causes marginal revenue per passenger as a result of an improvement inquality to drop (to the level of the regulated price), and the operator will nottherefore ®nd it pro®table to provide the same level of quality as it otherwise wouldhave done. Both quality subsidies and quality regulation leave marginal revenue perpassenger unaVected and these instruments therefore produce the same results.

5. DiscussionThis paper has analysed the consequences of entry-deterring strategies for the

provision of quality in the local bus market. It has examined a hypothetical busoperator on an urban bus route and used simulation to determine the strategies itmay pursue, the welfare impacts of those strategies and the impact regulation mayhave.

A pro®t-maximizing monopolist protected from entry will undersupply both inthe quality and in the frequency dimension. If the incumbent ®rm does face entrythreats, it will run more buses as to avoid leaving pro®table gaps. It will stillundersupply in the quality dimension, though it supplies more quality than a fullyprotected monopolist would do. This leads to small welfare gains compared with thefull monopoly position.

However, the welfare levels achieved come nowhere near the levels attainable andit has therefore been examined whether regulation or subsidy systems can deliver abetter outcome. Quality regulation and quality-related subsidy systems were eVectivein most cases, though only to a limited extent as prices would remain at themonopoly level. Substantial welfare gains can only be realized if prices are lowerthan that, which can be achieved either by price regulation or output subsidies.Unlike price regulation, output-related subsidies do not cause the monopolist to cutin the quality dimension. Subsidy systems will however increase the pro®tability ofthe ®rm and this may lead to even more oversupply in the frequency dimension. Toavoid this, lump-sum taxes should be levied, e.g. by franchising the service.

The results carry important implications for policymakers. Quality Partnershipscan be regarded as some hybrid between lump sum and quality-related subsidies.Both were seen to lead to welfare losses or, in case of quality-related subsidies, to anambiguous outcome. There is therefore a risk that Quality Partnerships lead towelfare losses if increased operator pro®tability causes even more oversupply in thefrequency dimension than may have been the case already. Protecting the qualityoperator from entry will not be a solution, as the operator will then adopt the fullmonopoly position, which is probably even worse than the deterrence equilibrium.Even if Quality Partnerships are accompanied by lump-sum taxation, the potentialwelfare gains remain limited as long as fares remain at the monopoly level. Priceregulation or, even better, output subsidies should therefore be given seriousconsideration.

It is worth noting the limitations of the work reported here. The model is a highlysimpli®ed version of reality, based on simulated data, and does not take thedistinction between peak and oV peak, any segmentation of the market or thepossibility of price discrimination into account, to mention just a few issues. Forexample, when including a possibility of diVerent fares for diVerent market segments,the consequences of monopoly pricing might be less severe. They would even beabsent if the operator is able to discriminate perfectly, but that is probably even less


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

realistic than the assumption in this model of a standard ¯at fare, at least in theurban bus market.

Another limitation is the fact that vehicle size in the model is ®xed, i.e. nodiVerent vehicle sizes can simultaneously operate in the model. If an incumbentoperates a ¯eet of full-sized buses, it will be able to deter operators entering withsimilarly sized vehicles but it might not be possible to deter minibus entry. On theother hand, an incumbent operator faced with such a situation might choose toswitch to minibus operation itself.

Furthermore, the decisions of the ®rm have only been analysed by looking at themarginal revenue and marginal cost of a change in quality. In practice, however,these are not the only factors the ®rm will take into account. There could be politicalfactors, as a result of which the ®rm might supply more quality than otherwise toavoid regulatory intervention. The ®rm may also use the quality instrument as abarrier to entry. If the market is too small to sustain more than one quality level, theincumbent ®rm could make entry unattractive by choosing a quality level that wouldinvolve high sunk costs for any entrant to match. In the extreme, this could lead to asituation where a ®rm oversupplies in the quality dimension, thereby deterring entry,and no longer needs to oversupply in the frequency dimension. The ®rm would bebetter oV if oversupplying quality is cheaper than oversupplying bus miles, but thewelfare implications could be serious. This, too, could be a consequence of QualityPartnerships. Finally, it should be mentioned that no estimate has been made of thepotential external welfare impacts of Quality Partnerships, for example throughreduced car use and therefore less congestion.

An interesting issue is the extent to which the results could be applied to a routewhere two operators are competing. If entry by the second operator does not resultin competition on price (as discussed in Section 2), the operators might reach acompetitive equilibrium where they deter entry by a third operator. This equilibriumcould have similar properties as the deterrence equilibriums discussed in the presentpaper. There might, however, be some complicating factors, such as the possibility ofirregular headways in a competitive situation.

Suggestions for further research include repeating the analysis using real-worlddata and better information on the costs and valuation of quality measures,whenever this becomes available. The case suggested above where the ®rm detersentry by oversupplying vertically as opposed to horizontally might be worthexploring as well. The analysis could also be extended by looking at quality provisionin more competitive markets, where prices might not be at the monopoly level orwhere two or more quality levels might co-exist. Furthermore, existing urbantransport models could be improved by distinguishing between the two qualitydimensions. The traditionally horizontally diVerentiated bus market has seen apolicy emphasis in recent years on vertical product diVerentiation and the previouslyunexplored case of a market where products are both horizontally and verticallydiVerentiated would therefore deserve more attention.

AcknowledgementsThe author thanks Professor Peter Mackie and Dr Jeremy Toner, University of

Leeds, for many useful discussions, as well as an anonymous referee for valuablecomments.


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

ReferencesABBOTT, L., 1953, Vertical equilibrium under pure quality competition. American Economic

Review, 43, 826 ± 845.CHARTERED INSTITUTE OF PUBLIC FINANCE AND ACCOUNTANCY, 1974, Passenger Transport

Operations (London: CIPFA).CUNDILL, M. A. and WATTS, P. F., 1973, Bus Boarding and Alighting Times. TRRL Report

521.EVANS, A. W., 1987, A theoretical comparison of competition with other economic regimes for

bus services. Journal of Transport Economics and Policy, 21, 7 ± 36.MACKIE, P. J. and PRESTON, J., 1996, The Local Bus Market (Aldershot: Avebury).SCHMALENSEE, R., 1978, Entry deterrence in the ready-to-eat breakfast cereal industry. Bell

Journal of Economics, 9, 305 ± 327.UK DEPARTMENT OF THE ENVIRONMENT, TRANSPORT AND THE REGIONS, 1999, From Workhorse

to Thoroughbred: A Better Role for Bus Travel (London: HMSO).UK DEPARTMENT OF THE ENVIRONMENT, TRANSPORT AND THE REGIONS 2000, Transport 2010:

The 10 Year Plan (London: HMSO).WHELAN, G. A., TONER, J. P., PRESTON, J. M. and SHIRES, J. D., 2000, Modelling quality bus

partnerships. Working paper, University of Leeds.


Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

3:03

16

Oct

ober

201

4

entry deterrence and quality provision in the local bus market

Documents