lable at ScienceDirect

Research in Transportation Economics 30 (2010) 59e73

Contents lists avai

Research in Transportation Economics

journal homepage: www.elsevier .com/locate /retrec

Applying social marginal cost pricing in rail PPPs: Present state, drawbacks andways forward

João Bernardino a,*, Zden�ek H�rebí�cek b, Carlos Marques c

a TIS.pt, Consultores em Transportes, Inovacao e Sistemas, S.A., Lisbon, Portugalb Transport Research Centre (CDV), Brno, Check Republicc Instituto Superior Técnico (IST), Universidade Técnica de Lisboa (UTL), Lisbon, Portugal

Keywords:Social marginal cost pricingPublic-private partnershipsRailways

* Corresponding author. Tel.: þ351 21 350 4400.E-mail address: joao.bernardino@tis.pt (J. Bernard

0739-8859/$ e see front matter � 2010 Elsevier Ltd.doi:10.1016/j.retrec.2010.10.008

a b s t r a c t

The application of social marginal cost pricing (SMCP) in PPP’s in the railway sector faces several chal-lenges. We examine in detail the practical applicability of SMCP in railway PPPs from the perspectives ofcost accounting and effectiveness of SMCP towards the allocative efficiency goal, addressing the likelydrawbacks in conciliating the welfare objectives of SMCP with the objectives of project financing (costrecovery) and value for money that justify the realization of PPP’s. To this end, we combine theoreticalanalysis with the observation of empirical results of a case study. We split the analysis per type of privateservice provision, which can be for service operation or infrastructure management. For infrastructuremanagement, we recommend splitting the operator remuneration and the track access charges. Forservice operation, we argue that the correct decision on source of funding of the service operator shoulddepend on the characteristics of the contract.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

The application of social marginal cost pricing (SMCP) in PPPs inthe railway sector faces several particular challenges, related withthe nature of rail marginal costs, with the nature of centralizedplanning and management of railway operations, and with thevarious existing scopes for private involvement in railway provi-sion. We analyse the most relevant issues and their possibleconsequences to conciliation of SMCP with PPPs, and try to identifypractical ways to adopt what should be feasible and effectiveapproaches to the application of SMCP.

We start by reproducing reporting on trends and concretecases of deployment of PPPs in the rail sector in Europe andproceed with a description of the social marginal costs involvedin this sector. Drawing from the particular features of railwaytransport, we then examine in detail the practical applicability ofSMCP from the perspectives of effectiveness of SMCP towards theallocative efficiency goal, addressing also the likely drawbacks inconciliating the welfare objectives of SMCP with the projectfinancing (cost recovery) and value for money objectives that

ino).

All rights reserved.

justify the use of PPPs. To this end, we combine a theoreticalanalysis with results of a case study of an urban railway linelinking the two sides of the Tagus River in the Lisbon area,realized within the ENACT1 study.

2. PPPs in the railway sector

2.1. Types of private involvement in railways

Railways were, in its early times, mainly constructed and oper-ated by the private sector. However, with time, it became clear thatnetwork economies and reduced scope for competition put railwaysin a situation where a pure market was not the most beneficialsystem, and States began to take over their construction and oper-ation. More recently, governments started increasingly relying onpartnerships with the private sector for railway projects andundertakings, using regulation to guarantee their appropriateperformance.

1 The ENACT study was commissioned by the European Commission under the6th Framework program. The study developed in the period 2006e2009 and wasled by Prof Rosário Macário (TIS.PT, Consultores em Transportes, Inovação eSistemas).

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e7360

Since the recent unbundling of infrastructure network operationand train service operation, private involvement in railway supplycan take two major forms:

� Service operation� Infrastructure management

By private involvement in service operation we simply meanthat the provision of railway services is performed by a privateentity, which may or not own the assets e the rolling stock. Infra-structure management involves the supply, maintenance and,possibly, investment in railway infrastructure.

The first type is regarded as the “competitive” portion of thesector, where different private operators are expected to competefor the provision of train services and compete for demand withinand outside the rail mode. This form of private involvement isalready commonly used, and its application at a successful rate hasbeen demonstrated under adequate regulatory frameworks.

It can be discussed whether private involvement in railwayservice provision can be regarded as a publiceprivate partnership.The term “Public-Private Partnership” (“PPP”) has been in generaluse since the 1990’s, yet there is no widely agreed single definitionor model of a PPP. In the 2003 European Commission Guidelines forsuccessful PublicePrivate Partnerships, the term of PPPwas broadlydefined as ”a partnership between the public sector and the privatesector for the purpose of delivering a project or a service tradi-tionally provided by the public sector”. General characteristics ofPPP arrangements that are proposed in the EC’s Green Paper onPPPs (COM (2004) 327 final) are the relatively long duration of therelationship, the method of funding the project - in part from theprivate sector - the important role of the economic operator, whoparticipates at different stages in the project, and the distribution ofrisks between the public partner and the private partner. By thisaccount, simple franchises of railway service operation could beregarded outside the PPP universe, since contract durations andinvestments by the private party can be small and its participationmay be restricted to a single activity (service operation). None-theless, private involvement in this area still poses importantquestions as to its conciliationwith SMCP, and therefore we includeservice operation in our object of analysis.

Infrastructure management has less scope for competition, andcertainly no significant scope for competition for demand. Still,a few PPPs for infrastructure have been carried out in Europe. Theycomprise both negative and positive outcomes, and the subject ofattribution of infrastructure management to a private party is stillsubject to controversy (see e.g. Pollit & Smith, 2001; Vickerman,2004). While the step back with network privatization in GreatBritain and failures in other sites represented a major drawback,there are also cases that seem successful, and there is a set of newrail infrastructure PPPs under preparation in Europe. It is stilluncertain whether this trend will prevail. For reference, in thefollowing section we present a set of full-scale, infrastructure buildand operate PPP experiences, in the European States.

2.2. Use of infrastructure PPPs in European states

2.2.1. Trends in the use of infrastructure PPPsAccording to the OECD report on transport infrastructure

investment (OECD, 2008), PPP projects of the type seen in the roadsector are less prevalent as far as rail infrastructure is concerned.This is perhaps due to the fact that rail is generally managed ona network basis, and because railway operators are already typi-cally at an arm’s length from government decision-making in manycountries, as a result of their organisational structures. Notwith-standing, in all monitored countries there is a growing number of

PPP rail projects in the preparatory or realization phase. In mostcases, these ventures provide a special service that is somehowdifferentiated from the rest of the network operation. In thefollowing pages we present some relevant examples.

2.2.2. Infrastructure PPPs in European States2.2.2.1. Austria. The example of PPP is the project Nordkettenbahn(NKB) in Innsbruck. The project is realized by Strabag (InnsbruckerNordkettenbahn GmbH). The provider of the concession is NKB; thedaughter company Strabag hands over economic risks and cashesfor the financing incomes. The total investment is 51 mil. EURO(ENACT, 2008a).

2.2.2.2. Great Britain. One example is the Channel Tunnel Rail Link(CTRL). CTRL was launched in 1993, as the largest project underprivate finance initiative, to connect London with the ChannelTunnel, and therefore speed up travel time to Paris on the Eurostar.The project was also an EU Trans-European Networks (TENs)project. Revenue forecasts proved to be highly optimistic, resultingin the British government having to backstop the concessionairewith a loan guarantee. On the basis of this guarantee, in 2006 theUK Office of National Statistics determined that the governmenthad a controlling interest over the project’s parent company,London & Continental Railways, which was thus reclassified asa “public non-financial corporation”.

In a similar vein, the Eurotunnel Group, created in 1986 to buildand operate the Channel Tunnel, has struggled with the initial debtincurred for the project, which cost six times more than initiallyprojected.

Another example of private finance initiative is the PPP toprovide, maintain, renew and upgrade elements of the LondonUnderground metro system. This involves 3 separate 30-yearcontracts for different elements of the work, with payments basedon performance, including bonuses for surpassing a given cap, andpenalties for not meeting it (OECD, 2008).

2.2.2.3. Czech Republic. Currently in Czech republic there is not anypractical experience with realized PPP projects in the field of therailway transport. One project is under preparation, covering theUpgrade of the Prague-Kladno railway line plus construction ofa railway connection to the Ruzyn�e Airport. It features a combina-tion of BOT/DBFO and operation and management contract, costingaround 500 million EURO and having a contract term of 30e40years. Meanwhile the emphasis has been placed above all on thelegislation treatment of the mentioned activities.

The following legal provisions are under question:

e proposal of Law about publiceprivate partnerships (franchiselaw);

e proposal of Law about public orders;e amendment of Law about estate of the congestion/scarcity and

its appearance in legal relationship;e proposal of Law about budgetary rules.

2.2.2.4. France. PPP funding is utilized in the railway transportinfrastructure in France. An example is the building and operationrailway Perpignan-Figueras Rail Concession. The main target of theproject is the reduction of time needed for the border crossingbetween the two states and the reduction of the railway operationon the French-Spanish border. The total investment is 1.1 thousandmillion EURO. The private party is responsible for building andoperation of the railway during the time of 50 years. The publicsector, which also includes the government of Spain, ensures theproject design and a partial funding of 540 million EURO. The

Table 1Comparison of unit Activity Cost Inputs (current prices).

UK Cost (e000s)

Rail renewal (e/mile) 254Sleeper renewal (e/mile) 486Ballast renewal (e/mile) 501Points renewal (e/mile) 181Tamping (e/mile) 6Rail repair (e/failure) 2Points repair (e/failure) 10

Sweden Cost (e)Track wear and tear (per gross km) 0,0003Use of marshalling yards (per wagon, occasion) 0,4457

Source: Railtrack and Swedish National Rail Administration (edited in ENACT,2008a).

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e73 61

railway construction has started in February 2004. The finalizationis planned in 2010.

2.2.2.5. Germany. Wemention two relevant projects in Germany inthe field of railway infrastructure. The first is the Freiberg-Holzhau(Erzgebirgsstrecke), where a 31 km infrastructure is operatedprivately since 2000, with a franchise agreement of 19 years andinvestment costs of 8.8 million EURO.

Another example is the Heidekrautbahn, where the privateinfrastructure operator manages a line of 34.4, with a franchiseagreement of 15 years and comprising investment costs of 17.3million EURO.

2.2.2.6. Italy. An example of PPP in Italy is the construction of LaNuova Linea Metropolitana di Milano M5 by the Municipality ofMilan. The PPP is a DBFO, with the start date in 2007, a buildingperiod of 5 years, investment costs amounted to 502 million EUROand a Partner Agreement of 32 years.

2.2.2.7. Netherlands. Another rail-based PPP, established in 1999, isthe Netherlands’ High-Speed Line (HSL) rail link between Antwerpand Amsterdam, based on a 30-year concession. The concessionaireis remunerated by government on the basis of a performanceagreement, which demands 99% compliance in order for the privatepartner to receive its full payment, and there was no transfer ofdemand risk (OECD, 2008).

2.2.2.8. Sweden. PPP financing in Sweden is used in the field of therailway infrastructure very often. For instance, we can mention theproject Arlanda. In return for paying for about 70% of the infra-structure investment, the concessionaire is entitled to charge trainpassengers both to pay for operating the train and in order torecover the initial investment, over a period of 45 years. TheArlanda contract assigns both market and cost risk to the conces-sionaire. The Swedish government provides a “guarantee loan” tothe operator which is subordinate to all private debt; by postponinginterest and debt retirement until private debt has been repaid, thisresults in a reduction of the company’s costs for debt service duringits first years of operation (OECD, 2008).

2.2.3. Prospects for infrastructure PPPsAs a conclusion, OECD (2008) remarks that these types of

examples largely demonstrate the fact that rail PPPs are likely tofocus on specific services or needs, as an adjunct to the largernetworks, which are also managed using devolution models, typi-cally state-run or private organisations.

3. Social marginal costs in the railway sector

3.1. Social marginal costs in the railway sector

The main barriers to the introduction of social marginal costpricing in rail infrastructure are related with the difficulties of itsmeasurement (especially for congestion and scarcity) and the factthat they do not provide the right incentive for investments,financing and the transformation of railways into marketableentities. Moreover, rail infrastructure managers have an aim to usecharging revenues to cover their total costs, or at least a greaterproportion of costs, than those covered through marginal costpricing. Rail cost categories and estimation methodologies arediscussed below.

3.1.1. Marginal cost of infrastructureAvailable infrastructure cost accounts for the rail sector distin-

guish usually between the cost of operation and maintenance,

administration and overheads and the capital costs. Amore detailedcost categorization is certainly available inside the rail companies.Furthermore, in the creditedebit balances costs are categorized.However, there is no official categorisation of variable and fixedcosts. It should also benoted that infrastructure cost accounts for therail sector have to deal with the problem of how to distinguishbetween the costs of tracks in the narrow sense and those relatedwith other facilities such as stations and freight terminals (HighLevel Group on Transport Infrastructure Charging, 1999a).

Moreover, the accounting systems currently applied by infra-structure managers have not been designed with the scope ofsetting charges. Therefore, a transition is needed towards moreelaborated accounting systems able to provide the adequate inputfor cost models (activity based costing, life cycle costing, etc.) andcharging calculation procedures (marginal cost calculation)(RAILCALC, 2007). Studies for estimating marginal infrastructurecosts have been carried out in Germany and Sweden. DB estimatesthat 15%e20% of costs are variable costs. The Swedish analysis ofmarginal maintenance costs (Joansson & Nilsson, 2002) hasempirically shown that 10% of the average costs for maintenancearemarginal costs, which implies that the share of marginal costs inall average costs is even lower (Table 1) .

Railtrack, the previous owner of UK railway infrastructure, hasundertaken specific research for the structure of rail access charges.Within this study asset usage costs (marginal costs of maintainingand renewing assets) were estimated and indicated that theproportion of Railtrack’s costs that vary with asset usage is around10e15% (High Level Group on Transport Infrastructure Charging,1999a).

GRACE project case studies (Lindberg, 2006) performed esti-mations on marginal costs in the rail sector. Estimations indicated0.00070 €/ton-km in Sweden for renewal and maintenance and0.00097 €/ton-km in Switzerland. Maintenance only has a cost of0.00031 €/ton-km in Sweden and 0.00045 €/ton-km inSwitzerland. The marginal cost in UK is estimated in 0.002 €/ton-km. Operation has a marginal cost of 0.054 €/train-km in Sweden.The Hungarian study concludes on a marginal cost of 0.22 €/train-km for train movements.

When short-run marginal costs are calculated through econo-metric analysis, the cost categories involved are operation, main-tenance and renewal (or one or two among them). It can be arguedthatmost of the current experiences calculating short-runmarginalcosts through econometric models sum up maintenance andrenewal costs as a single dependent variable. This practise isgenerally justified by the short series of data currently availablethat could be distorted by the long periodicity related to renewals(RAILCALC, 2007).

NERA (1998) gives some guidelines about which cost elementsare relevant for deriving short-run marginal costs. These costsrelate to an existing non-congested rail network, and contain:

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e7362

e Additional track wear and tear;e Traction current;e Signal operation costs;e Train planning costs;e Management and administration costs;e Costs of disruption caused to other train services.

3.1.2. Marginal cost of congestion and scarcityExternal costs in rail are generally small, except possibly those of

scarcity costs, which may be important depending on infrastruc-ture capacity and demand. Existing evidence shows that scarcity isindirectly included in some existing pricing schemes by consid-ering demand or quality level of tracks or slots.

Estimation of congestion costs in rail is more complex than isthe case on roads or airports, due to complexity caused by networksand schedules.

As reported by IMPACT (2007), the UNITE study suggestsmarginal external congestion figures in rail transport is around 20€/train-km in morning peak based on UK and Swiss evidence.However, the positive externality of additional demand (Mohringeffect) based on Swedish data is believed to be 5e10 times higherthan external rail congestion costs. Such low congestion costsestimates are in line with the findings of the COMPETE countryresults, stating that congestion in rail transport is often eliminatedby respective adoptions of time tables.

The High Level Group on Infrastructure Charging (1999b) andothers suggest to better use the value that train operators put onthe scarcity of tracks as a measure of infrastructure capacity utili-zation. In other words, scarcity costs should be valued by theopportunity cost of the slot in question. Slot auctioning is thusa possibility for valuing scarcity costs, but it may be rendereddifficult due to the complexities involved in terms of the alternativeways in which the infrastructure may be used. Complementarily toscarcity costs, it is suggested that unscheduled delays imposed byone train operator on another may be charged ex post.

A different approach, recommended by NERA (1998), is to iden-tify sections of infrastructure where capacity is constrained and tocharge the long run average incremental cost of expanding capacity,although this is a very difficult concept to measure (Nash andMatthews, 2002). An alternative approach considered in GRACE isfor the track charging authority to attempt to calculate directly thecosts involved in depriving another operator of the slot, for instancethrough the value people place on departure time shifts. The GRACEcase study on the stretch of the East CoastMain Line from London toDoncaster shows that scarce tracks in peak hours might be over 10times more valuable than tracks in off-peak (Lindberg, 2006).

Only Great Britain has a congestion charge per train-kmexplicitly related to estimates of congestion costs. However,charges per train-km in Italy and Germany vary by train speed andtype of route. In Germany there is an explicit utilization factor, witha higher charge for heavily used lines. Italy uses a simple approachof setting standard speed profiles for each route designed to opti-mise the line, and charging higher prices for paths that deviatefrom the profile, either by seeking faster or slower paths thatdisrupt the optimal service profile. Slovenia is proposing an off-

Table 2Unit values for noise marginal costs for different network types of rail traffic - (Eurocent

Type of train Time of day Urban

Passenger Day 23.65 (23.65e46.73Night 77.99

Freight Day 41.93 (41.93e101.1Night 171.06

Central values in bold.Source: ENACT (2008a), data aggregated from IMPACT, 2007.

peak discount. There is also a charge per node in Switzerland andItaly that varies with the implicit amount of congestion at the nodeby categorizing nodes according to traffic levels (EuropeanConference of Ministers of Transport, 2005).

3.1.3. Marginal costs of environmentAir pollution and climate change costs of trains are negligible

compared to infrastructure costs, except for diesel trains. Airpollution costs in €/train-km passenger and freight trains (fromHEATCO and CAFE CBA cost factors for Germany used) varybetween around 0,1 and 0,6 €/train-km for electric trains and 1,4 to7,5 for diesel trains. Climate change costs vary between 0,03€/train-km and 0,55 €/train-km, while diesel trains vary from 0,12to 0,62 €/train-km (IMPACT, 2007).

Noise emissions are dependent on train speed, the coach/wagontype, surface conditions of both wheel and rail, and type of track(including the level of maintenance). Closely related to these arecost drivers like the type of brakes, the length of the train, and thepresence of noise walls (ENACT, 2007). Recommended outputvalues for rail noise are presented in Table 2.

3.1.4. Marginal cost of accidentsFew studies on marginal (or rather average) accident costs exist

for railways. In INFRAS/IWW (2004) average external accidentcosts for rail transport are calculated based on up-to-date UICaccident statistics. The Swiss case study within UNITE (UNITE,2002) also presents values for average external accident costs forrail transport. Costs for accidents between rail and other modesare allocated based on a causation perspective (and thereforemostly attributed to road transport). European average externalaccident costs for rail transport amount to 0.08e0.30 €/train-km(IMPACT, 2007).

3.2. Are current charging practises in Europe compliant with SMCP?

Previous chapters of this special issue describe current chargingpractises throughout European countries, including the railwaysector. As it shows, current infrastructure use charging practisesrelate only partly with the principles of SMCP, with significantdifferences from country to country.

Marginal cost pricing is advocated with the aim of encouragingefficient use of the railwaynetwork.However railways tend to exhibiteconomies of density and consequently the marginal cost of extranetwork utilization is below the average cost. Thus, full cost recoverycannot be achieved through (simple) marginal cost pricing. Twoalternative pricing principles are applied: marginal cost pricing withmark-up (MCþ) and the pricing to recover full cost less governmentgrants (FC-)which is applied only inGermanyand Italy, aswewill seein the following section. Both MCþ and FC- are aimed at full costrecovery less government grants; however theMCþ approach, beingbased onmarginal cost pricing, is viewed as less distorting in terms ofincentives. Maintenance, renewals, train planning and operations,congestion and scarcity, accidents and environmental costs are usedas the cost base to determine both social marginal cost and averagecost. No country charges for all of these categories but all countries,

/vkm).

Suburban Rural

) 20,61(10.43e20.61) 2,57 (1.30e2.57)34.40 4.29

7) 40.06 (20.68e40.06) 5.00 (2.58e5.00)67.71 8.45

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e73 63

except Italy, charge for maintenance expenditure. Some even chargefor trainplanningandoperations. Charges forcongestionandscarcity,accidents and environment are only undertaken by a minority ofcountries. In particular, environmental/accident costs are included inthe chargingmethodologyonly in Sweden,while scarcity/congestioncosts are taken into account in France and Germany (DIFFERENT,2006).

The RAILCALC (2007) study assessed the extent to which currentaccounting and charging practises are aligned with the objectivesand requirements mentioned in Directive 2001/14/EC on the allo-cation of railway infrastructure capacity and the levying of chargesfor the use of railway infrastructure (European Commission, 2001).The analysis has specifically focused on the aspects of cost recovery,cost relatedness, incentives to cost reduction, differentiationaccording to services, differentiation according to type of line,consideration of demand’s willingness to pay, incentives to optimaluse of capacity and complexity.

The analytical procedure has been applied to various chargingelements identified as being used by railway infrastructuremanagers, including cost of use of assets, mark-ups, reservationcharges, performance schemes, congestion charges, scarcitycharges, environmental charges and subsidies and discounts. Thefollowing table (Table 3) synthesises the confrontation of cost of useof assets (CUA), which are the predominant marginal costs forrailway infrastructure costs, with the objectives and requirementsof the Directive. The study identified fifteen different basic chargingschemes related to the cost of use of assets practised throughoutEuropean countries.

Part of the basic charging schemes identified was assessed ashaving a positive alignment with SMCP related objectives andrequirements, although most had a negative judgement. Similaranalyses were performed for the remaining identified chargingelements used across Europe. Besides charges related to costs of useof assets, the other types of charges found and analysed weremark-ups, reservation charges, performance schemes, congestioncharges, scarcity charges, environmental charges and subsidies anddiscounts, with differing results. An important observation of theproject was that the conciliation of all objectives and requirementsforeseen by legislation was not fully possible.

4. Applying SMCP on privately provided railways e fromtheory to practice

4.1. Why may SMCP interfere with private involvement in railwayprovision?

Like in other transport modes, social marginal cost pricing inrailways may interfere with private involvement in railway provi-sion because it imposes a price, and the imposed price may notfunction appropriately from various perspectives:

Table 3Confrontation of cost of use of assets with objectives and requirements of Directive 2011

CUA 01 02 03

Economic efficiency þ þ þNon-discriminatory access þ þ þIncentives to cost efficiency � � �Cost-relatedness (CUA) þ þ þDifferentation according to specificities � � �Uniformity of charging principles þ þ þNon discriminatory charges for different RU þ þ þMarginal cost approach þ þ þRelation between charges and costs attributable to infra services � � þ

Practices codes indicated in x5.8: (þ) Positive alignment (�) Negative alignment.Source: RAILCALC (2007).

� SMC price may not generate enough revenues to pay theprivate operator

� SMC price formation may introduce high risks of revenue, andthereby become an impediment to private participation undera reasonable compensation

� SMC price formation may introduce undesirable incentives onthe performance of the private operator

These types of constraints have specificities in railways incomparison with other transport sectors, specially due to differ-ences in price structure and dynamics. They also differ with thetype of activity attributed to the private party.

Interference is only important if the operator is remuneratedthrough user charges. In face of the potential constraints imposed bySMCP on private involvement in railway provision, it becomes clearthat important problems are only caused when the operator is paidon the basis of user charges. Otherwise, SMCP will not have directinfluence neither on the level of payments nor on revenue risks orincentives to the private party (for a detailed presentation of theargument, see ENACT, 2008b).

Exception to this rule may be if the private operator is remu-nerated proportionally to demand (through a shadow toll), a case inwhich the price indirectly affects the operator’s funding by influ-encing the demand level. Still, in this case, the effects should besmall and not necessarily undesirable.

Of course, the possibility of no interference of SMCPwith privateinvolvement by splitting charging revenues from operator'sremuneration does not mean that undesirable financing, risk andincentive problems may not occur. They are, instead, transferred tothe public party. But this situation is not different in essence froma situation where there is no private involvement, where revenuedrawbacks from SMCP affect the public party anyway.

As such, how may SMCP affect private involvement in each type ofactivity? As we have seen above, private involvement in railwayprovision may include two distinct types of activity: infrastructuremanagement and service operation. The way in which SMCP mayaffect private involvement has different incidence in the twoactivities, depending on funding mechanisms used and on SMCcharge formation considerations for each activity.

Section 4.2 describes the differences in framework on fundingand charging for the two types of activity e infrastructuremanagement and service operation. The following sections exploremore precisely how SMCP may affect private involvement, andwhich ways forward are available to conciliate SMCP and privateinterests in the involvement in railway operation.

Section 4.3 identifies the factors of interference between SMCPand private involvement, including financing, risks and incentives.Section 4.4 analyses these factors in more detail, in the scope of therailway sector. Section 4.5 discusses possible ways forward toconciliate SMCP with railway PPPs (separately for service operationand infrastructure management), including analysis on different

/14/EC.

04 05 06 07 08 09 10 11 12 13 14 15

þ þ þ � � � � � � � � �þ þ þ þ þ þ þ þ þ þ þ þþ � þ � þ � � þ þ � � �þ þ þ � � � � � � � � �� þ � � þ þ � þ þ þ � þþ þ þ þ � þ þ þ þ þ þ þþ þ þ þ þ þ þ þ þ þ þ þþ þ þ � � � � � � � � �þ þ þ � þ � þ þ þ þ � �

Fig. 1. Interaction between SMC price and decisions on railway service/infrastructureprovision.

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e7364

final user price formation schemes, identification of usablecontractual performance incentives and assessment of their need,and a final discussion on who should be the owner of SMC pricingrevenues.

4.2. Funding and charging frameworks

4.2.1. Should railway service users be charged according to SMCP?At least since the talks of the EC High Level Group on Transport

Infrastructure Charging (1999a), the framework of analysis onsocial marginal cost pricing in all modes has been focused on infra-structure charging, rather than charging for all infrastructure andoperation related costs. However, a recurrent theme of the discus-sions in the group was the different nature of ‘the customer’ treatedby a road infrastructure owner and a rail infrastructure owner(Goodwin, 2001). Road customers consist of atomistic users - indi-viduals or firms, while rail customers, following separation of trackand operations, are a small number of operating companies, so thatthe relationshipwith the final customer (i.e. the passengers) is moredistant.

In this scope, should rail passengers be charged according toSMCP (and how)? The issue hasn’t been discussed in depth becausethe terms of reference of the Group focused explicitly on infra-structure charging, rather than charging for all infrastructure andoperations. Yet, the question is of crucial relevancy in the rail sector,not only for the issue of how to correctly introduce SMCP but alsofor its implications towards conciliation of SMCP with privateinvolvement in rail service operation.

Social marginal costs are, in all sectors, essentially vehicledependent, i.e. they vary with features associated to the vehicle,namely its characteristics, distances covered, time of day of oper-ation or the possible presence of other vehicles in the infrastruc-ture. The amount of passengers or goods inside the vehicles is fairlyirrelevant for the amount of SMCs caused, except to some extent inthe use of stations. For these reasons, the internalization of costsmust primarily be targeted on vehicle operation rather thanpassenger or goods size. Such premise implies that, for all transportsectors, it should be vehicles, not passengers, to be priced for theuse of the infrastructure.

On the other hand, the aim of optimizing social behaviourbehind the principle of social marginal cost pricing should hold thatfinal users of the transport service bear the social marginal costs oftheir travel decisions. This implies that users of the service shouldbe priced in accordance to social marginal costs. However, whereasin the scope of private transport this is straightforwardly done dueto the non-existence of an intermediary service operator, in time-tabled transport the application of pure passenger short-run socialmarginal costs does not make users face the marginal costs of thevehicle operation because vehicle circulations will occur, in theshort-run, independently of the users decisions.

However, for dynamic efficiency of provision and use of trans-port services, it seems that the more appropriate alternative is tomake the user pay for a due share of the marginal costs caused bythe vehicle operation. A traveller, in choosing between road andrail, should in principle consider the social cost implied by bothmodes, even if the train circulation will happen independently ofthe traveller’s instantaneous decision on the mode to use. It is theSMC price inclusive of vehicle marginal costs that brings an optimaldemand distribution between the various modes. And it is thisexpected optimal demand that should, in the first place, base themedium-term decision to have trains circulating as established bythe decision maker (Fig. 1).

But will an optimal service user price (SMC based) occur if it isthe service operator being directly charged for track accessaccording to SMCP? According to economic theory, in a competitive

market, user prices would be set by the operator equal to themarginal costs of the service operated. Therefore, if service opera-tors were charged for track access according to SMC, then the costwould be perfectly transferred to the final users. This is presumablythe case of unregulated sectors like roads and aviation, wheretransport service providers (buses and airlines) are allowed tofreely set prices to passengers on the basis of the assumption thatmarkets are competitive. But if the service operator has marketpower and is able to determine user prices, the intended effects ofoperator SMC charging will be distorted, a case in which therewould have to be regulatory intervention to set user prices equal toSMC.

In the railway sector it is likely that in most cases, especially inthe passenger market, the specific features of the railway service e

limited capacity, network coverage, comfort, speed, and the exis-tence of captives (passengers without personal vehicle, goods oflarge size and weight) - cannot be fully replaced by the availablecompetitors. Therefore, in a large piece of railway undertakings, thecase cannot be made that the market is competitive, and serviceuser charges should be imposed to reflect social marginal costs.

As mentioned, past discussions and research on social marginalcost pricing in the railway sector have been focused on infra-structure management. The issue of marginal costs of serviceoperation in railways is therefore a subject deserving furtherresearch in the scope of SMC pricing. To this aim, two elementsshould be covered:

� Which types of service operation costs should be regarded asmarginal costs?

� Quantification of marginal service operation costs.

The first item deals with an issue already subject to discussion inthe scope of railway infrastructure costs, where it is not alwaysevident which costs should be considered as variable and whichshould be considered fixed (see e.g. Rothengatter, 2003). Theframework of analysis must in any case be laid in comarison withothermodes; similar types of costs should have the same treatmentin different modes, for SMCP to provide the right incentives (fora discussion of policy packaging, se e.g. Verhoef, 2001). This is ofparamount importance in considering marginal rail service oper-ation costs; as argued, final user price should include this type ofcosts despite the fact that they occur independently of

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e73 65

instantaneous final user decisions, so that demand adjusts effi-ciently to the alternative transport supply alternatives.

For example, it is not clear whether costs of train acquisitionshould be considered as marginal costs in rail, when comparingwith other modes. Whereas they are not included in the traveldecision equation of road users (and consequently they should notbe covered by the rail price), on the contrary they are included inprices charged to air transport passengers.

We may thus conclude that, unless service operation costs areincluded properly in final users prices throughout all sectors, therewill be a distortion in final user choices.

4.2.2. Possible funding and charging frameworks per type of activityTo assess possible constraints put by SMCP on private involve-

ment in railway provision, we need to understand the funding andcharging framework associated with the possible activities ofprivate involvement e infrastructure management and serviceoperation.

According to the principle of economic efficiency under thepresence of external costs, it is efficient that the party causing theexternal cost compensates the party affected for the causedburden.

Applying this rule in infrastructure management PPPs wouldimply that the users of the infrastructure pay the affected partiesfor the social marginal costs of their activity in the infrastructure. Inthis sense it would be appropriate for the infrastructure user to paythe external costs of infrastructure to the infrastructure managerand the external costs e congestion/scarcity, environmental, acci-dent e to society.

However, in the context of PPPs it would simplify the process oftransference of funds, and perhaps also incentive good practice bythe manager by having it facing commercial demand risks, if theinfrastructure manager were entitled to the payments for externalcosts. In this way, the external costs caused by the infrastructureusers would be internalized and the State would be partly or totallyexcluded from funding the infrastructure manager, as expressed inFig. 2. The approach of funding the infrastructure manager throughtrack access charges is indeed used by some of the existing infra-structure PPPs in railways (KuhlI, 2007).

Service operation requires a different approach. Should it be theservice operator paying for social marginal costs to the infrastruc-turemanager and/or the State, or should those be borne by the finalusers of the transport service?

Fig. 2. Possible infrastructure management remuneration under an SMCP scheme(users pay access charges e SMCs e fully to infrastructure manager).

Like in infrastructure management PPPs, in service operationPPPs the above stated principle of economic efficiency and fairnessunder externalities should hold that each party pay the other forthe costs inflicted. In this sense the service operator should pay theinfrastructure manager for infrastructure marginal costs caused bythe service, and the State/society for the other social marginal costs.In addition, as argued above, for the sake of transport user behav-iour optimization, the final users of the service should pay theservice operator not only for the short-run marginal costs causedby themselves (which probably are minimal) but also for the socialmarginal costs incurred by the operated service.

Similarly to infrastructure management PPPs, the process offinancial flows could be simplified if the service operator fullykept the tariffs paid by the users (instead of directing socialmarginal costs to the State) as form of compensation for theiractivity (Fig. 3).

4.3. Factors of interference between SMCP and private involvement

The factors of interference between SMC pricing objectives andthe objectives behind the involvement of the private sector inrailways are generally of the same kind as other transport sectors:

� Cost coverage: The revenues generated by user charges basedon SMCs may not be sufficient to cover for the necessaryremuneration of the private operator

� Risks of revenue: The formation of revenues from SMCP mayintroduce high risks of revenue

� Incentives: The formation of revenues from SMCP may intro-duce undesirable incentives on the performance of the privateoperator

4.3.1. Cost coverageThe problem of financing the infrastructure or transport

undertaking is invoked as the coremotive behind the late growth inthe use of PPPs in Europe. This seems to be true also in the railsector. In conventional PPPs of concession type, the private operatoris entitled to the funds generated by the exploitation of the infra-structure or service. However, the imposition of a price at the SMClevel may make this solution impracticable, as the generatedrevenues may not be sufficient to cover all the costs.

4.3.2. Risks of revenueThe possible introduction of SMCP may have consequences on

financial risks of transport projects, due to the potential variabilityof SMCs. In conventional pricing schemes with fixed price, the riskassociatedwith revenues fromuser charges is directly related to theuncertainty of future demand. But when marginal cost basedpricing is introduced, two additional risk factors are introduced:

� Demand based revenue risk: risk associated with the possiblevariation of price with demand, caused in particular bycongestion/scarcity costs;

� Future evolution of SMCs: risk associated with the externaluncertainty of evolution of SMCs.

The possible introduction of additional revenue risks will betranslated into an additional risk premium for the operator to bewilling to bear them,whichmay be unacceptable from a perspectiveof value for money for the public sector.

4.3.3. IncentivesThe difference in objectives of the public and private parties in

a PPP arrangement puts a challenge on the conciliation of those

Fig. 3. Service operation concessionaire remuneration under an SMCP scheme (users pay access charges e SMCs e fully to service operator; service operator pays infrastructurecharges to infrastructure manager).

Table 4Revenues generated by social marginal cost pricing in the Tagus Crossing service.

Cost category Average cost Revenues total (PV)

(€/train.km) (€*1000) (%)

1. Infrastructure 1,29 48.628 77,0%2. Congestion/Scarcity 0,20 7.471 11,8%3. Air Pollution 0,04 1.373 2,2%

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e7366

objectives. The introduction of a price based on social marginalcosts may even aggravate the difficulty of this challenge. If theprivate operator has powers to influence social marginal costs and,in that way, revenues, it will be in its interest to see SMCs at a highlevel.

At a theoretical level, one way SMCs may change is throughchanges in the congestion/scarcity ratio, affecting the way how thedemand incentive works. In fact, the possibility of increasing reve-nues through capacity restrictions, by increasing congestion/scar-city costs, may lead the operator to prefer providing a reduced levelof service, even at the expense of a lower demand.

Besides capacity provision, the operator has an interest in seeingSMCs at a high level through any other possible means. Therefore, ifits powers allow it in some way to influence the real or apparentlevel of SMCs, it is expectable that it will try to influence themupwards. For example, if price has a component proportional to theaccident risk, it will be a profitable interest of the operator to seethat risk increase.

4.4. In practise, how may SMCP interfere with private involvement?

Cost coverage, risks and incentives are the main economicobstacles to the conciliation of SMCP with PPPs. In this topic weaddress how these factors act, in particular, in the scope of railways.Additionally, we also analyse non-economic factors of potentialrelevancy to the issue in the railway sector, namely acceptabilityissues and the need for contract renegotiations.

Due to the differences in funding and charging approaches inthe two types of activity where private involvement is possible, weseparately cover service operation and infrastructure managementPPPs. SMCP has direct implication for the private involvement inrailway provision for the purpose of:

e Service operation, if the funding of the service operator is basedon SMCP charges for the use of the service or if it is SMCPcharged for track access.

e Infrastructure management, if the funding of the infrastructuremanager is based on track access charges built on SMCP;

4. Global Warming 0,04 1.697 2,7%5. Noise 0,10 3.691 5,8%6. Accident 0,01 285 0,5%Total 1,68 63.145 100,0%

Source: ENACT Case Study B - Tagus River Rail Crossing (Bernardino, 2009).

4.4.1. Cost coverageCost coverage in the scope of service operation PPPs focuses on

whether the revenues delivered by passenger or goods charges

would be enough to pay for all the costs incurred by the privateoperator.

4.4.1.1. Service operation. One of the ENACT case studies analysedthe performance of a service operation PPP (the river Tagus crossingsuburban railway line in Lisbon) (Bernardino, 2009). It has beenestimated that the internalization of the marginal costs typicallyconsidered (IMPACT, 2007) for the use of infrastructure would notlead to cost coverage of the service operation (as previouslyconcluded in previous studies, as noted e.g. by Nash and Matthews,2002). In fact, the SMC price would stay considerably belowcurrently practised passenger prices (Table 4).

The results show that revenues from social marginal cost pricingwould rely fundamentally on infrastructure costs (77.7%). Consid-ering the aggregate revenues of infrastructure, congestion andnoise costs, they represent 94.7% of all SMCP revenues. This result isa consequence of the well known good performance of railwaytransport in the environmental and safety dimensions. It must alsobe noted that both air pollution costs and global warming costswould not be considered in SMC price if they were considered toalready being internalized in other forms of taxation (particularlya fuel tax or an emissions trading scheme).

However, as we have seen above, the proper reflection ofmarginal costs on passenger or goods tariffs in the scope of serviceoperation should include the consideration of operational marginalcosts of service provision, which were not accounted for in the casestudy. Their inclusion in the passenger price could change the costcoverage picture, depending on the types of service operation costsconsidered as marginal.

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e73 67

4.4.1.2. Infrastructure management. Aswe have seen above (Section3.1.1), railway infrastructure provision has very high fixed costscompared to variable costs of its operation, and consequentlymarginal costs lie far below its average cost. This implies thata deficit occurs if price is set equal to infrastructure marginal costs.Furthermore, there are economies of traffic density, which meanthat marginal infrastructure costs decrease as traffic increases.

Although considering external costs from the use of infrastruc-ture e in particular, congestion, air and noise pollution, and acci-dents e in the price provides additional revenues, they will still befar below cost recovery of infrastructure costs. As we can see in theprice calculation of the ENACT Tagus River Rail Crossing Case Study(Table 5), external costs addminor additional revenues (about 50%).Even under much higher congestion/scarcity costs than thoseconsidered, cost recovery in rail infrastructure would still be farfrom achievable.

However, it is noteworthy that new railway projects in generalhardly manage to gather enough funds to pay for service operationcosts alone. As a result, any surplus revenues (net of service oper-ation costs) drawn from current user charges at most are able toprovide a small contribution to cover investment and other fixedinfrastructure costs. Whereas, in the road sector, PPPs are used bysome governments with the main purpose of avoiding financialburden to the public budget, the same is not generally possible inrailway projects to the extent that revenues are far from enough topay for all costs involved. The obvious conclusion is that the situ-ation on cost recovery of fixed infrastructure costs would not suffera very significant relative change by the introduction of SMCP,comparing to the present state.

4.4.2. RisksHaving the concessionaire’s remuneration based on SMC pricing

revenues has implications on the risks faced through the amount ofrevenues collected. As described above, two main topics arise inthis scope:

� Effects of demand risk on revenue risk (demand based risk)� Effects of evolution of SMCs over time on revenues (risk ofevolution of SMCs)

4.4.2.1. Demand based risk. Unlike in conventional demand basedrevenue schemes where revenues relate linearly with demand,this is not necessarily the case when revenues are dependent notonly on demand but also on the tariffs imposed in accordance withsocial marginal costs; tariffs based on pure SMCs may, on theirown, depend on demand and circulation conditions in non-linearways. Consequently, whereas in conventional demand basedrevenues the expectations of revenue risk are equal to theexpectations of demand risk, in SMCP based revenues this may notbe the case.

In service operation, because vehicle demand (in what theconsidered railway service is concerned) is in most cases pre-determined in contract, the effect of vehicle demand variations

Table 5SMC passenger tariff formation alternatives.

SMC passenger tariff formation alternatives

1. Actual SMC þ Actual number of Users Tij ¼ SMCajNj

2. Actual SMC þ Fixed Expected number of users Tij ¼ SMCajEðNjÞ

3. Expected SMC þ Actual number of Users Tij ¼ EðSMCjÞNj

4. Expected SMC þ Fixed Expected number of users Tij ¼ EðSMCjÞEðNjÞ

Source: ENACT Case Study B - Tagus River Rail Crossing (Bernardino, 2009).

over variations of social marginal costs may be neglected as anissue. In fact, the total amount of social marginal costs that arecaused by the train circulations of the PPP arrangement is in someextent pre-determined - in what dues to the quantity of demand -by the decision of vehicle demand in the infrastructure. Such is thecase e.g. of the Tagus river rail concession, where train circulationsfor the urban service are pre-determined in contract. Therefore, theissue of the risk of (vehicle) demand is not relevant in the scope ofcentrally determined timetabled transport service franchises.

This apparently would largely simplify the issue of risks ofdemand based revenues. However, unlike in infrastructure PPPs, inservice operation PPPs revenues are determined on the basis ofpassenger or goods demand, not vehicle demand. It is the fees paidby the users of the service that will supply revenues. This intro-duces a different source of risk, analysed in detail in Section 4.5.2.

In infrastructure management PPPs, demand based risk couldtheoretically be aggravated by the existence of SMC pricing, if theinfrastructure suffers from congestion and scarcity. However, thislater provision may not be the case in most of the existent orplanned PPPs, which, as we have seen in Section 2, mostly focus ondedicated lines for specific services that are less likely to facerelevant congestion and scarcity problems, which tend to occur inmore interlinked networks. However, if congestion/scarcity isrelevant in the infrastructure, an additional revenue risk caused bythe pricing scheme would surge.

Due to their lower relevancy compared to other modes, infra-structure management PPPs in the rail sector have not been studiedin detail in ENACT. However, the applied theoretical propositions onSMCP revenue risks and incentives are similar for all modes. Toillustrate the possible influence of the demand/capacity relation onprice, and consequently on risks and incentives, we replicate inFig. 3 the conceptual functional form relating price for congestion/scarcity and the demand/capacity relation, as applied in the ENACTSimulation Tool for non-road modes.

The reasoning behind the curve is based on the differentcomponents that affect congestion and scarcity costs in scheduledtransport:

e For low levels of demand (compared to capacity), there are nocongestion/scarcity costs;

e For higher levels of demand,marginal costs increase as demandapproaches capacity due to the need to adjust schedules toaccommodate all vehicles and to the increased risks of propa-gation of delays;

e When potential demand2 is higher than capacity, in addition tothe costs mentioned in the previous case, opportunity costsarise from the non-satisfaction of part of the potential demand.

The concave form of the relation between demand and priceresults in that possible variations in demand (or capacity) translateinto more than linearly proportional variations in price. Conse-quently, the revenue risk related to demand variations is higherwith SMCP with congestion/scarcity costs included than witha pricing scheme where price does not depend on demand.

4.4.2.2. Risk of evolution of SMCs. If pure SMC pricing is applied,revenues depend on the evolution of social marginal costs duringthe time horizon of the PPP, which may be uncertain.

The SMC price would, as we have seen, be formed largely byinfrastructure costs, with possibly some importance of conges-tion/scarcity costs, and little influence of other external costs. The

2 Potential demand is only satisfied if it is lower than (maximum) capacity.Otherwise, an excess demand is not satisfied.

Fig. 4. Functional form used in the ENACT Simulation Tool for SMCs of scarcity/congestion (SMCs/c) as a function of the demand/capacity ratio, for non-road modes.

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e7368

contribution to risks of revenue from costs other than infra-structure or congestion/scarcity can therefore be regarded asnegligible.

As for infrastructure costs, they may change in relation to theinitial expectations either due to technological modifications ininfrastructure or rolling stock, or due to operational efficiency ofthe infrastructure manager. The rolling stock in the railway sector isless subject to technological evolution as compared e.g. with theroad sector and therefore future SMCs should be more easilypredictable.

Congestion/scarcity costs could evolve mainly due to timetablechanges in the network towards a more or less busy network, ordue to technological changes in network operation. In any of thecases, these may represent a risk of tariff revenues due to theirimplications on the SMC price practised.

Such risk may be more or less relevant depending on the PPPcontract extension. The longer it is, the more risk of evolution ofsocial marginal costs there will be. In the case of service operationfranchises, risks should be small, since contracts last usually formore limited periods with close to fixed technology.

4.4.3. IncentivesThe theoretical ways in which SMCP can negatively affect the

incentives faced by the private party were categorized in ENACT intwo groups: incentives to influence revenues through capacityprovision; incentives to keep SMCs at a high level. It is now relevantto consider how these incentives would function in the railwaysector, both in the scope of service operation and infrastructuremanagement.

In service operation, like for risks, incentives caused by demand/capacity relationmay likely be neglected inmost situations becausethe demand is generally pre-determined by contract. Where thisisn’t true, i.e. the operator has power to decide on servicefrequencies, the allocation of social marginal costs to the final user(passengers or goods) could bring an undesirable incentive to theoperator to increase frequencies and with that seeing congestion/scarcity cost revenues increase.

Railway service operation PPPs have particular kinds of incen-tives to increase SMCs, not always similar to those incentives foundin infrastructure management PPP’s. The types of incentives thatarise in the scope of service operation PPP’s are the following:

� Demand incentive - The incentive of the operator to increasepassenger demand leads it to putting a higher effort to providea service of high quality, including possibly the increase ofservice frequency (if that is an option to the operator). Thisincentive may or not be affected by SMCP, depending on priceformation rules, as we will see in the following section.

� High externality incentive - the private party has an interest intaking actions to keeping marginal costs at a high level, in caseits powers allow it to do so. In railway services the operatormay e.g. increase the number of carriages to increase infra-structure costs, underperform on schedule to increasecongestion costs, or underperform on safety actions.

� Innovation and cost efficiency incentive e in the scope of serviceoperation PPP’s and SMCP, the incentive for innovation andcost efficiency may be affected if improvements in cost effi-ciency of the operator translate into correspondent losses inrevenues. This may be the case if the SMC price includesa fraction for service operation costs (not considered inprevious Chapter 4), since the service operator will not have aninterest in decreasing them if they are not translated intoadditional profits.

� Information asymmetry incentive e an incentive to hide infor-mation from the regulator on real costs would happen, as

above, if service operation costs took part in the SMC price. Inthat case, the service operator would have an interest inreporting higher costs than the real ones in order to chargea higher price. Another possible action could be hiding infor-mation on energy consumptionwhere, like above, the operatorwould have an interest to report higher consumption than real.

The three later incentives have in common the interest of theconcessionaire to raise the SMC price. All of themmay be negativelyaffected by SMCP. The demand incentive, on the contrary, pushesthe operator to increase revenues through an increase in demand.

The way in which SMCP affects the incentives described isgreatly affected by the rules of price formation, as we will seebelow.

In infrastructure management PPPs, the demand/capacityincentive will only be relevant, likewise for risk, if there is signifi-cant congestion and scarcity, which as mentioned above should notbe the case in most existent or planned rail infrastructure PPPs. Yet,if that is the case, an undesirable incentive to limit capacity may bereal and should have to be neutralized. As is clear from the obser-vation of Fig. 4, a decrease of capacity could be of interest to theprivate infrastructure operator, since the increase of the demand/capacity would increase the price faced by railway undertakings.

Like in other transport modes, other undesirable incentives mayoccur in the scope of infrastructure operation. The demand incen-tive will be negatively affected under congestion/scarcity costs,given that the operator may be willing to lose demand by capacityrestrictions aimed at increasing SMC price. The high externalityincentive could be relevant in the rail infrastructure especially inthe scope of accident costs. The innovation and cost efficiencyincentive should be of paramount importance due to the weight ofinfrastructure costs in the SMC price, and the power of the infra-structure operator to influence those costs through time. If it is paidaccording to infrastructure costs, the incentive to minimize costsand innovate will be null or even negative. The possible informationasymmetry between the infrastructure manager and the regulatorcould also lead the biased reporting of costs, particularly byinflating marginal infrastructure costs. Additionally to the incen-tives referred for service operation, an additional undesirableincentive may surge in the scope of infrastructure operation:

� User discrimination incentive - if the concessionaire has theability to choose between users, it will have an incentive tochoose those users with higher marginal costs. The access torailway infrastructure may be excludable, especially if there isany scarcity; therefore, the operator may discriminate userswith lower costs to users with higher costs.

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e73 69

4.4.4. Other barriers of non-economic natureThis topic looks briefly into other possible barriers, of non-

economic nature, to the conciliation of SMCP with privateinvolvement in railways. In the railway sector, we highlight thepossible importance of public acceptability and the renegotiation ofcontracts.

4.4.4.1. Public acceptability. The transfer of common pricingschemes to SMC pricing could theoretically bring problems ofpublic acceptability, either due to the level of pricing or due to theuse of SMCP revenues. However, neither of the issues seems topotentially bring any significant problems in the scope of railways.The new prices will, in principle, be lower than current chargingpractises. And the issue of the revenue destination does not seem tobe a major potential constraint for acceptability of SMCP imple-mentation in railways, since presently user charging revenuesalready commonly fully flow to the operators (public or private).

4.4.4.2. Renegotiation of contracts. Usually, concession contractsbetween the State and private parties establish that the renego-tiation of contractual conditions requires the agreement of bothparties. In the case of a possible renegotiation launched by theState with the aim of introducing SMCP, the predictable rationalbehaviour of the private party is the attempt to reap benefits outof the renegotiation. In fact, the circumstance of the private partyhaving the power to veto the renegotiation intended by the publicparty may lead it to strategic behaviour in the pursuit of a bettercontract.

This kind of practise may be aggravated by the lower objectivityimplied by SMCP in the definition of revenues based on userpricing. Particularly, the possible introduction of the additionalrevenue risks of SMCP could provide the private party with a strongargument for demanding a generous contract renegotiation for itsside.

However, the problem can be potentially neutralized in twoways. First, SMC pricing could be established in a non-risky way, i.e.such that the prices would be predictable and not subject toconditions of demand or evolution of social marginal costs throughtime (e.g. due to technological improvements). As described below(section 4.5), this seems to be viable in timetabled service opera-tion, but not in infrastructure management.

The second possibility is to simply substitute any previousrevenues from user charges by correspondingly equal performancepayments proportional to demand. In this case, the public partywould be the one collecting revenues from SMCP user charges. Theonly subsisting additional risk faced under this situation would bethe risk of demand introduced by variations in user charges,a possibility that again may be neutralized if SMC charges aredefined in a predictable way.

Finally, it should be noted that the problem of need for negoti-ations tends to be less serious in railway service operation than ininfrastructure management PPP’s, since contracts have commonlya smaller term, possibly allowing for its termination before theintroduction of SMCP.

4.5. Possible ways forward to conciliate SMCP with railway PPPs

Possible ways forward to the conciliation of SMCP and railwayPPPs are covered here. We describe solutions for complementaryor alternative funding, analyse the way how alternative possibleprice formation schemes can be important in the creation ofinconsistencies between SMCP and private provision, identifyrequirements and possible solutions for contractual performancedrivers under different pricing and remuneration regimes, andfinally we address the issue of who should be the party entitled to

collect user charge revenues, from an optimal contractual designperspective.

4.5.1. FundingIt was seen that internalization of the marginal costs typically

considered for the use of infrastructure would not lead to costcoverage of service operation or infrastructure build & operatePPPs. In fact, the SMC price would stay considerably belowcurrently practised passenger prices, meaning that deficits wouldincrease with SMCP in relation to the present practise. However,the proper reflection of marginal costs on passenger or goodstariffs in the scope of service operation should include theconsideration of marginal costs of service operation. Their inclu-sion in the passenger price could change the cost coverage picture,depending on the types of service operation costs considered asmarginal.

Like in other transport sectors, additional funding wouldrequire the use of alternative means of revenue additionally toSMC user pricing, and the main alternatives available are:

� Additional public funding;� Second-best pricing schemes to achieve more revenues.

Additional public funding enables applying first-best SMCP, butpresses on public deficits. Second-best pricing schemes prejudiceallocative efficiency in the mobility market and are more complexto implement. These issues are discussed in previous chapters ofthis book and ENACT (2007).

As to what the railway sector is concerned, a few points can bemade:

4.5.1.1. Competition. Second-best pricing schemes get morecomplex to implement when there is competition between variousmodes. Railway lines are in most cases in direct competition withother modes, mostly road, but also other railway services, aviation(for high-speed rail) and even inland waterways. Competition ismost deep in urban areas. Therefore, the application of second-bestpricing in railway lines will generally require complex schemes inintegration with other modes.

4.5.1.2. Deficit level. The deficit in the railway sector as a whole isthe highest among transport modes. Second-best pricing mayaggravate the picture, as a result of lower prices on final users. Still,some service operation PPPs might be financially viable throughSMCP if marginal costs of service operation are considered.However, the issue of operational marginal costs of service provi-sion in railways is a subject deserving further research in the scopeof SMCP.

4.5.1.3. Cross-subsidization. According to Roy (2002) the surplusesresulting from the correction of urban road congestion far outweighany deficits resulting from the application of marginal cost pricingelsewhere in the transport system. Therefore, it is possible toremunerate portions of the transport system in deficit, includingrailways, through surplus SMCP revenues in urban roads. Such typeof solution would dispense any additional public funding orincreased pricing.

4.5.2. Price formation in service operation4.5.2.1. Price formation alternatives. The risks and incentivescaused by SMCP in the context of service operation depend largelyon the rules of formation of passenger tariffs.

According to the principle, suggested above, that final users ofthe transport service should face a price that reflects the socialmarginal costs involved in the railway service, the sum of the tariffs

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e7370

paid by the users should equal the social marginal costs of thevehicle operation:

XN

i

Tij ¼ SMCj

With Tij e tariff of user I in vehicle-trip jSMCj e social marginal costs of vehicle-trip j

The application of this rule still allows for different alternativesof user tariff formation schemes contractually agreed between thepublic and private parties of the PPP. Tariff formation alternativescould differ in two ways:

� Tariffs may be alternatively set according to expected or actualsocial marginal costs. Setting tariffs in accordance with actualSMC would consist on the regular updating of tariffs in accor-dance to the evolution of the level of SMC, whichmay vary withtime due to changes in technology, flow conditions or otherfactors. In alternative, tariffs may be set according to expectedSMC by a fixed value, defined at contract agreement and basedon the SMC expected to happen during the contract period.

� Tariffs may be alternatively set according to the expected oractual number of users,with tariff formation similar to the above.

The combinations of the possibilities outlined lead to fourpractical alternatives of passenger tariff formation, within a PPPcontract. They are presented in (Table 5).

4.5.2.2. Performance of price formation alternatives. The advantagesand disadvantages of these alternatives on risks and incentives, andon market efficiency, may be analysed.

Focussing on the effects of the passenger tariff alternatives onrisks and behaviour incentives to the party collecting the revenues,a qualitative analysis is provided and outlined in (Table 6).

In relation to number of passengers, tariffs based on actualdemand (i.e. real time demand)would neutralize both revenue risksand incentives, due to the fact that the tariff adapts permanently toproduce revenues equal to social marginal costs and thereforerevenues are independent of passenger numbers. It is noteworthy tomention that the neutralization of risks and incentives is of absolutenature, in the sense that besidesnot producing any incremental risksand incentives in relation to conventional demand based revenueschemes, it actually totally neutralizes them, turning out to bea similar remuneration scheme to a fixed subsidy from the State. Inopposition, a tariff based on the initially expected demand would

Table 6Effects of alternative passenger tariff formation rules on risks and incentives.

SMC passenger tariff formationalternatives

Effects towards:

Revenue Risks BehaviourIncentives

1. Actual SMC Actual Number of Users Caused by:- evolution ofSMC’s

No incentives

2. Actual SMC Expected Number ofUsers

Caused by:- evolution ofSMC’s- passengerdemand

- raise SMC’s (�)- increase users(þ)

3. Expected SMC Actual Number ofUsers

No revenue risks No incentives

4. Expected SMC Expected Number ofUsers

Caused by:- passengerdemand

- increase users(þ)

Legend: (þ) - desirable incentive ; (�) - undesirable incentive.Source: ENACT Case Study B - Tagus River Rail Crossing (Bernardino, 2009).

produce risks and incentives, namely a risk of revenues dependenton the demand and an incentive to increase the number of passen-gers. The later case is, in what directly depends on passengerdemand, no different from those of conventional demand basedrevenue schemes. It would require a possible dependence betweenpassenger demand and socialmarginal costs to introduce additionalindirect effects of passenger demand over risks and incentives;however, as noted above, passenger numbers hardly affect the socialmarginal costs of a scheduled service.

In the case of tariffs being formed on the basis of expectednumber of passengers, it can be foreseen that having passengertariffs formed on the basis of actual social marginal costs (i.e.permanently adapting tariffs to the current value of social marginalcosts) would put a risk on revenues and an incentive on the oper-ator to increase the social marginal costs of its operation, which isan undesirable outcome. If, instead, the tariff formation scheme isbased on the initially expected SMC’s during the whole course ofthe scheduled service, then no revenue risks or undesirableincentives will be caused by possible SMC variations. In this respect,the renegotiation of the Tagus rail crossing would easily enablea tariff formation scheme based on expected marginal costsbecause it was set for a period of only 6 years, allowing for anaccurate prediction of marginal costs during that time.

Evaluating the available tariff formation alternatives from thescope of risks and incentives introduced, it can be observed thatalternatives 1 and 2 are dominated by alternatives 3 and 4, i.e. thelater alternatives show a better or at least equal performance in thetwo dimensions of analysis. Comparing the “winner” alternatives, 3and 4, it can be seen that there is a trade-off between risk bearingand productive efficiency. While alternative 3 has no risks andincentives whatsoever, alternative 4 introduces a desirable incen-tive on quality of service at the cost of an additional risk premiumdue to revenue uncertainty.

Moreover, analysing the performance of the two “winner”alternatives, it is doubtful that the one relying on actual demandwould work well from the practical perspective given the unde-sirable feedback price/demand inter-relations that may surge. E.g.a lower than expected level of demand would lead to an increase ofpassenger prices, which would again lead to a further decrease indemand, and so on. This price setting scheme may thus introduceinstability in the price/demand equilibrium which may evendamage the original objectives of SMCP. For this reason, the alter-native based on expected costs and users (alternative 4) seems tobe the best choice of SMCP tariff setting rules.

Tariff formation rules based on expected costs instead of actualcosts is a deviation frompure SMCP, if costs do not evolve as expectedduring the period concerned. Nonetheless, in the case of a PPP inwhich train circulations aredeterminedbycontract (asusual), andaretherefore relatively inflexible during that period, the costs caused bytrain circulations do not depend on the tariff level (which determinesthe number of passengers). What is crucial for an efficient allocationof resources is that the initial decision on the volume of train circu-lations takes into account theirexpected socialmarginal costs and theexpected passenger demand with SMC pricing. After the medium-term decision is taken, it seems to be reasonably irrelevant fordynamicefficiency topriceusersaccording to theexpectedSMC’s thatbased the decision or the SMC’s that truly occur. For this reason,possible market inefficiencies caused by deviations from pure SMCPin theshort term(withfixed train circulations) shouldnotbe relevant.However, this plausible hypothesis could be a subject of furtherresearch.

The solution of having SMC prices formed on the basis ofexpected rather than actual costs seems, for the reasons pointedabove, to be more viable in the railway sector than in the roadsector. On one hand, its timetabled nature removes some

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e73 71

undesirable risks and incentives and, on the other hand, there isgreater predictability of technological elements.

4.5.3. Performance incentivesContractual design is key to achieving a desirable performance

of the private operator. The requirements of performance drivers ina PPP with SMCP depend on:

1. The revenue structure of the operator (which could or not bepartly based on user tariffs);

2. The rules of formation of price.

If revenues of the operator are based on other sources thanpassenger tariffs, the incentives it faces are independent of thepricing regime. If, on the other hand, the operator owns revenuesfrom passenger tariffs, the incentive effects of SMCP depend on thetariff formation rules, as described above. In the later situation,there may be the need to counterbalance negative incentivesproduced by SMCP. This would be the case of incentives wherebythe operator has advantages in contributing to an increase in SMCprice (demand/capacity incentive, high externality incentive,innovation and cost efficiency incentive, information asymmetryincentive and user discrimination incentive).

Answers for these undesirable incentives are provided in greatextent by types of performance drivers already applied in presentrailway service operation contracts. For example, the high exter-nality incentive can be addressed by penalties related to delays,passenger safety, maintenance of material goods and adequacy ofthe transport capacity to the demand levels. The problem ofinformation asymmetry is also commonly addressed by penaltiesalready foreseen in service operation contracts like the one coveredin the ENACT case study. The problem of innovation and cost effi-ciency incentive can be solved through price setting rules, bysetting price according to initially expected costs instead of actualcosts; as described above (Section 4.5.3), this solution wouldeliminate any incentives to raise SMCs. An alternativemechanism ishaving a form of benchmark regulation whereby the operatorwould be awarded or penalized for its cost efficiency compared toa given benchmark.

Table 7 provides a qualitative account of the different require-ments for performance drivers, according to pricing scheme, allo-cation of revenues and price setting rules.

A conventional pricing scheme with tariff revenues owned bythe operator is one with a comparatively low need for performancedrivers because it features a demand incentive which pushes theoperator to provide a high quality service in order to attractdemand.

Under an SMCP scheme, the need for performance driversdepends on the owner of the tariff revenues and varies with the

Table 7Level of need for contractual performance drivers and monitoring, depending on the pri

Conventional pricing sc(actual PPP)

Incentives Demand (þ)

Requirements for performance drivers and monitoring þLegend: (þ) - desirable incentive ; (�) - undesirable incentive.Source: Adapted from ENACT Case Study B e Tagus River Rail Crossing (Bernardino, 200

rules of price setting. If the operator is remunerated through publicfunds (and not tariff revenues), the requirement is high comparedto the previous case, because in this situation there is nota demand incentive. In the situation where the operator is remu-nerated through tariff revenues, and price is set according to actualdemand, any incentives will be eliminated; this corresponds to anincentive situation similar to the one where the operator is fundedthrough public funds. In the case where price is set according toexpected demand and actual SMCs, there will be a demandincentive but also incentives for the operator to raise SMC’s. Due tothe later incentive, this solution should require performancedrivers of magnitude certainly higher than the conventionalpricing scheme standard, and perhaps even higher than the publicfunds solution. An alternative solution with price based onexpected SMCs would maintain the demand incentive and elimi-nate the SMC incentive, which puts it in a similar situation toa conventional pricing scheme, where the need for performancedrivers and monitoring is minimized in comparison with the otheralternatives.

In summary, the level of needs for performance incentives andmonitoring would be higher or equal than that under a conven-tional pricing scheme, depending on the revenue structure of theoperator and, in case it is based on user tariffs, on the rules of pricesetting. An SMCP based remuneration solution potentially offersless need for contractual performance drivers, since it may be ableto include a natural demand incentive.

4.5.4. Who should be the owner of user charge revenues?Perhaps the biggest choice to make for PPP contractual design is

to define how the private party is remunerated. In the traditionalconcessionmodel of PPPs, the State gives the private party the rightto exploit the infrastructure or service, collecting the revenuesgenerated by it. But, when prices are set on the basis of marginalcosts, the possible added risks and incentives are an additionalmajor issue to consider under this contractual choice problem.

In fact, the issues that arise concerning the willingness ofa private party to enter a PPP agreement and to provide sociallydesirable deeds by the existence of SMC pricing, are only relevant ifthe private party is remunerated through user charge revenues.

Like above, the type of analysis changes with the type of railwayactivity with private provision, i.e. onwhether we are talking aboutservice operation or infrastructure management.

4.5.4.1. Service operation. The analysis of the factors for anappropriate retention of SMCP revenues in timetabled railwayservice operation suggests that it would be feasible and, possibly,the best option, to maintain the status quo of concession typepartnerships and keep handing user charge revenues to the privateparty.

cing scheme.

SMC Pricing

heme private party’s revenues based on SMCP public funds

price setting based on . demand

actual expected

price setting based on . SMC’s

actual expected

Demand (þ)Raise SMC (�)

Demand (þ)

þþ þþ/þþþ þ þþ

9).

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e7372

An appropriate tariff setting scheme theoretically can simulta-neously fit the SMCP principle for railway passengers and maintainthe advantages of a conventional pricing scheme, which provide anincentive for good performance through attraction of endogenousdemand - avoiding additional needs for contractual performancedrivers. This conclusion departs from the assumption that finalusers should be the ultimate bearers of the marginal costs causedby the operation of a train, even though marginal costs are rela-tively independent of the number of passengers. Anotherassumption is that vehicle demand (train circulations) on theinfrastructure is determined in the beginning of the contract, andthat therefore congestion/scarcity costs of the train service wouldbe set a priori and not subject to changes during the contracttimeframe. The alternative price formation rules allowed by theabove principle have, as we have seen, different implications on therevenue risk and incentive framework; the analysis of the effects ofthe available price setting rule alternatives on risks, incentives andalso price/demand stability, resulted in that the alternative ofhaving the price formed on the basis of expected SMCs and numberof users would be the best. Such price setting alternative maintainsthe risk and incentive framework of a conventional pricing scheme,while guaranteeing dynamic stability of price and demand.Although such a tariff setting solution does not continuously followthe evolution of SMCs, it has still been considered an acceptablemarginal cost based price setting from the perspective of mobilitymarket efficiency, since SMC’s are determined a priori throughcentralized decision-making. The tariff setting scheme suggestedwould thus allow maintaining the status quo of toll based fundingpractises without affecting risks and incentives. Finally, the possi-bility of keeping current funding solutions in the railway sectorbased on user charging revenue should not carry significant addi-tional non-economic barriers, such as public acceptability.

However, public funding is also an option to consider. It mayeven be the most appropriate, depending on the degree of flexi-bility needed for price changes during the private granting time-frame. The possible need to allow price flexibility during thetimeframe of the contract, in order to adapt the price to anychanging circumstances, may justify the use of public funding.There are three ways why this need may arise:

1. There may be changes in the pricing framework of the trans-port system during the contract, with the need to adapt(second-best) pricing to implementation of marginal costbased pricing in other pieces of infrastructure;

2. Demand of train circulations may not be entirely determineda priori, with the service operator or the infrastructuremanager having the possibility of making changes duringcontract timeframe, and;

3. There may be changes in technological elements of infra-structure or rolling stock during the contract timeframe thatjustify the change of prices.

In either case, the traditional concession model may be aban-doned to a private remuneration scheme not directly related to usercharge revenues.

4.5.4.2. Infrastructure management. In infrastructure management,the need for price flexibility is imperative. Contracts for infrastruc-tureprovisions last for the long term,wheredemand, technologyandtransport network-price development are not predictable. The risksand incentives that thus takeplaceareof similarvein to thoseofothertransport infrastructures.

ENACT (2009) argued that a public funding based privateremuneration scheme would be generally most appropriate in theroad sector, whereas in airports, on the contrary, a user charge

remuneration scheme was regarded as a possibly preferable solu-tion. In the road sector, revenue risks and undesirable incentiveswere estimated high due to various factors, and their elimination isfeasible by removing user charge revenues from the private fund-ing. Moreover, the public funding solution is already widely appliedin motorways around Europe. On the contrary, in airports theremoval of user charges from the infrastructure operator’s remu-neration scheme was deemed difficult due to institutional accept-ability issues (see also the Chapter on airports in this special issue).In commercial aviation, a self-finance paradigm is dominant, andchanging it would likely face political and institutional barriers. Onthe other hand, current pricing schemes in airports usually alreadytake into account scarcity costs, and therefore the introduction ofpure SMCP would not produce totally new effects.

Railway infrastructure management seems to be closer to thecase of roads than airports. There is nothing like a financial self-sustainability paradigm in railways. Although pricing might takeaccount of scarcity costs in some railway networks in a few Euro-pean countries, a consistent market for railway service provision isnot yet established in the majority of cases. A pricing system basedon SMCP featuring highly variable scarcity costs is yet a relativelynew thing, and the use of PPPs is still immature. Moreover, a largepart of the railway infrastructure PPPs in development alreadyrelies on a private remuneration scheme based on performancepayments rather than user charging (KuhlI, 2007), which suggeststhat some regulators are already considering such a remunerationsolution as the most appropriate one, evenwithout the existence ofpure SMC pricing. For all these reasons, it seems that the publicfunding solution should be, in most cases, the best solution forprivate remuneration.

5. Conclusions

Private involvement in the provision of railway services hasbeen increasing in Europe, and is today extensive in service oper-ation and giving apparently growing steps in the scope of infra-structure management (although focused on specific services orneeds). Like in other transport modes, difficulties in conciliatingsocial marginal cost pricing with the use of publiceprivate part-nerships in the rail sector may arise.

Past discussions and research on SMCP in the railway sectorhave been focused on infrastructure management. We have arguedthat, to appropriately pursue the objective of allocative efficiency ofsocial marginal cost pricing, the discussion should be extended toservice operation. Final user (passengers or goods) pricing shouldinclude not only the marginal costs commonly considered forinfrastructure use but also those marginal costs directly related totrain operation, such that prices are comparable with other modes.Exactly which service operation costs should be considered asmarginal is something in need of further research.

Concerning cost recovery in the scope of infrastructuremanagement, present day revenues from user charging are ingeneral not enough to cover any significant part of the fixed costsinvolved in the investment on construction and management ofnew rail links. As such, a possible introduction of social marginalcost pricing should not cause a major change in the current situa-tion, in relative terms.

In the scope of service operation the implications are different,since rail service operator revenues presently extensively cover itscosts. It is clear that a final user (passenger or goods) price basedsolely on infrastructure costs and the remaining external costswould be rather lower than present tariffs. Yet, as mentioned above,operational marginal costs related to the rail service provisionshould also be included in the final user price, and their nature andamount is still unclear.

J. Bernardino et al. / Research in Transportation Economics 30 (2010) 59e73 73

Undesirable risks and incentives to the private party are causedby social marginal cost pricing, if its revenues rely on user charges.The problems introduced seem to be more important in the scopeof infrastructure management than in service operation, mainlydue to a difference in their typical contractual timeframes ofoperation and consequent on the rigidity of prices during theperiod.

In service operation, the analysis of the factors for an appro-priate owner of SMCP revenues in timetabled railway serviceoperation suggests that it would be feasible, and possibly the bestoption, to maintain the status quo of concession type partnershipsand keeping handing user charge revenues to the private party. Anappropriate tariff setting scheme theoretically can simultaneouslyfit the SMCP principle for railway passengers and maintain theadvantages of a conventional pricing scheme, which provides anincentive for good performance through attraction of endogenousdemand. However, a public funding solution may alternatively bethe appropriate option if, for any of a set of identified reasons, priceflexibility is required during the contract timeframe.

In infrastructure management, the need for price flexibility isimperative. Contracts for infrastructure provision last for the longterm, where demand, technology and transport network-pricedevelopments are not predictable. The risks and incentives thatthus take place are of similar vein to those of other transportinfrastructures. Moreover, unlike the airport sector, the railwaysector does not face any kind of institutionalized self-financingparadigm. Therefore, it is recommendable that the private infra-structure manager is compensated through means other than usercharge revenues.

References

Bernardino, J. (2009). ENACT case study B - Tagus rail crossing. ENACT- DesignAppropriate Contractual Relationships, Deliverable 5.1, FP6, EuropeanCommission.

DIFFERENT. (2006). Deliverable 7.1: Summary of selected practise for charging in therail sector. (User Reaction and Efficient Differentiation of Charges and Tolls,European Commission).

ENACT. (2007). Deliverable 1 e ENACT detailed methodology. TIS.PT. FP6, DG-TREN,European Commission.

ENACT. (2008a). Deliverable 2 - Social marginal cost pricing and second-best alter-natives, PART II-Infrastructure pricing schemes overview by mode. TRT. FP6, DG-TREN, European Commission.

ENACT. (2008b). Deliverable 3 e Incentive and contract theory. Rebel Group. FP6, DG-TREN, European Commission.

ENACT. (2009). Deliverable 5.1 - Case studies and Simulations - Comparison of resultsand interpretation of findings. TIS.PT. FP6, DG-TREN, European Commission.

European Commission. (2001). Directive 2001/14/EC of the European Parliamentand of the Council of 26 February 2001-on the allocation of railway infra-structure capacity and the levying of charges for the use of railway infra-structure and safety certification. Official Journal of the European Communities.15.3.2001, L 75/29.

European Conference of Ministers of Transport. (2005). Railway reform & charges forthe use of infrastructure. Paris: OECD.

Goodwin, P. (2001) What are the arguments really about? - Transport pricing andbroader economic and environmental objectives. IMPRINT-EUROPE Seminar,21e22 November 2001.

High Level Group on Transport Infrastructure Charging. (1999a). Calculating trans-port infrastructure costs. Final Report of the expert advisors (Working Group 1),April 28, 1999.

High Level Group on Transport Infrastructure Charging. (1999b) Final report onestimating transport costs. DG VII, European Commission.

IMPACT. (2007). Deliverable 1: Handbook on estimation of external costs in thetransport sector. Internalisation Measures and Policies for All External Cost ofTransport. European Commission.

Joansson, P., & Nilsson, J. E. (2002). Infrastructure cost case studies, Annex A3: Aneconomic analysis of track maintenance costs (UNITE). Funded by 5th FrameworkRTD Programme. ITS. Leeds: University of Leeds.

KuhlI. (2007). Rail Infrastructure projects e An European overview of implementationmodels, problems and shortcomings, 1999a. presentation held at the RailwayBusiness Forum, Warsaw, July 12th 2007.

Lindberg, G. (2006). Marginal cost case studies for road and rail transport, DeliverableD3, GRACE. Funded by Sixth Framework Programme. FP6, EuropeanCommission.

Nash, C., & Matthews, B. (2002). Implementing rail infrastructure charging reform -barriers and possible means of overcoming them. essay prepared for the secondseminar of the IMPRINT-EUROPE Thematic Network, Brussels, 14th/15th May2002.

NERA. (1998). An Examination of rail infrastructure charges. Final Report for theEuropean Commission, DG VII. May 1998, London.

OECD. (2008). Transport Infrastructure Investment: Option for Efficiency. OECDFinance & Investment/Insurance & Pensions, Vol. 2008(1). February 2008, pp.i-238(239).

Pollit, M. G., & Smith, A. S. J. (2001). The Restructuring and Privatisation of British Rail:Was it really that bad? DAE Working Paper 0118.

RAILCALC. (2007). “D-4: current practise assessment”. Supervisor: P. FonsecaTeixeira. In D. Fernandéz Belmonte (Ed.), Calculation of charges for the use of railinfrastructure. DG TREN, European Commission.

Rothengatter, W. (2003). How good is first-best? Marginal cost and other pricingprinciples for user charging in transport. Transport Policy, 10, 121e130.

UNITE. (2002). D10: Case studies on marginal infrastructure costs. DG TREN, EuropeanCommission.

Verhoef, E. T. (2001). Marginal cost based pricing in Transport: Key implementationissues from the economic perspective. Paper presented at the first IMPRINT-EUROPE seminar, Brussels, 21e22 November 2001.

Vickerman, R. (2004). Maintenance incentives under different infrastructureregimes. Utilities Policy, 12, 315e322.