Talgo Shunt

Arriva Celta

Arriva Spain Rail’s announcement of a new cross-border railway service from A Coruña (La Coruña) in Galicia to Porto (Oporto) in northern Portugal took some in the railway industry by surprise. The first proper phase of the liberalisation of Spain’s national passenger railways was widely expected to be focused on the high speed AVE network, a somewhat commercial near-aviation market, theoretically serviceable with trains acquired outside Spain. Even interest in cross-border services had hitherto focused on the high speed route from Madrid via Barcelona to the south of France, which judging by its latest search for 15 new cross-border drivers, state operator Renfe intends to respond to competitively. After all, the Spanish government had declared every regional railway service to be “Obligación de Servicio Público” (Public Service Obligation, OSP), to be financially supported as a Renfe monopoly, likely well into the 2030s. Add the difficulty of acquiring and operating uniquely Iberian gauged and signalled rolling stock in an environment where almost all the relevant assets are held by state operators, and one might dismiss the whole A Coruña-Porto scheme as an ill-conceived dream of a multinational that had not yet understood the local railway environment. Except the Arriva Group have been operating buses in Galicia since 1999 and Portugal since 2000, and so should know the territory as well as anyone. Perhaps more importantly, while Arriva’s British rivals sought liberalised markets for their initial forays “overseas” in the 1990s, Arriva learnt to work with whatever competitive environment it found on mainland Europe. That combination of local experience and competitive adaptability makes Arriva’s approach to Spanish railways unique. That Arriva’s first instinct is A Coruña-Porto, and not head-to-head competition on flagship intercity routes such as Madrid-Barcelona, reveals much about Spanish railway liberalisation. As explored in the following paragraphs, Arriva’s competitive strategy is contextualised by the need to:

  1. Address an underlying commercial market, not the “railway” market Renfe is structured around.
  2. Expose Renfe as no longer a “national” entity, thereafter making its role contestable.
  3. Exploit the structural weakness between national Renfe and regional government.

Arriva’s current parent group, Deutsche Bahn (DB), has already been burnt by the (non) liberalisation of Spanish railfreight: DB was fined 10.5 million euros by the Spanish competition authority, CNMC, for anti-competitive agreements to control the supply of traction – at rail-freight liberalisation in 2005, only Renfe Mercancías (its freight division) and DB subsidiary Transfesa (which had, exceptionally, operated privately since 1943 on RENFE‘s network) had freight locomotives that could operate across Spain. The case was emblematic of Renfe Mercancías’ anti-liberalisation strategy of asset control – from a non-auction of life-expired rolling stock that was all but useless for new operators, to a driver training paralysis that resulted in Renfe hiring competitors’ drivers (Renfe’s employment offer more favourable because of its public status). Renfe Mercancías’ approach was exceedingly traditional – defending the body of workers using the assets of the company – with scant regard for what those workers and assets were functionally there to do – move goods: Between its regular state-subsidised operating loses and its minimal impact on the national economy, Renfe Mercancías arguably lost track of its own importance and risks irrelevance. Yet even now the railway has few commercial logistics-orientated operators, and practical asset liberalisation is still dependent on the regulator forcing Renfe’s hand – hardly the panacea of free market liberalism envisaged by European Union policy. Even if this were the beginning of the end for Renfe Mercancías, it might also be the beginning of the end for freight on Spain’s railways: The underlying model (from the Spanish, not European Union, perspective) assumes the mercantile elite test the resolve of the national railway – the railway as a strategic entity of the nation – and only once that resolve has been broken does the national railway structure begin to yield actual liberalisation. However as simple transport utility, Spain’s internal freight was already somewhat liberalised on the roads, which convey the vast majority of Spain’s domestic goods traffic. Consequently it may have been easier to test the resolve of the national railway by capturing its freight market from the road, than compete on the railway itself: Renfe is easily blind-sided by functional competition that focuses on the market for moving goods, since Renfe’s organisational focus is “trains”, especially the staffing of trains.

While the problems of freight liberalisation have ensured intending commercial passenger railway operators are better prepared, for example establishing their own driver training schools and acquiring their own rolling stock, the underlying structure of Spain’s passenger railway liberalisation remains rather similar to that of freight. The baseline assumption, if only by historical pattern, is the emergence of a dominant commercial duopoly, plus Renfe: Perhaps a market similar to Spanish terrestrial television, where two dominant but counter-balancing commercial broadcasters co-exist with a state broadcaster. However, the dominant state actor is surely ADIF, the state owned provider of railway infrastructure: Commercial liberalised operators will rationally focus only on passenger “utility” (delivering a functional transport service), so, as discussed in The Expectations of Competition, the only unfilled role is that of “presence” (the physical manifestation of authority). While the original unified RENFE provided both utility and presence, on the modern railway presence can effectively be achieved by ADIF alone building and maintaining railway infrastructure: Exactly who operates the trains would not matter strategically, only that they were operated by someone. That cripples Renfe (specifically Renfe Operadora, its passenger division) as the strategic national entity it was, and explains Renfe’s desire to merge back with ADIF and thus remain part of a true national entity. Assuming Spain does not follow Poland and merge its national operator and infrastructure company back together, it will be incumbent on commercial operators to break Renfe’s monopoly position on track, so that Renfe can no longer be considered a national entity, thereafter allowing other operators to take Renfe’s role in subsequent non-commercial (typically regional and local) liberalisations. Mere dilution of Renfe’s patronage by indirect competition, such as by alternative modes, will not break Renfe as a national entity. Even competition only on high speed routes risks fostering a liberalised environment that cannot subsequently transfer to older (primarily) Iberian gauge networks, thus cannot compete effectively in whatever contractual regime eventually emerges for state-supported regional and local services, and hence maintains Renfe as a monopoly operator of much of the Spanish railway network.

If Renfe’s status as a national entity has been admonished by the time its OSPs are due for review (a maximum of 10 years after they are awarded), it will surely be politically impossible for Spain’s national government to continue to determine regional railway services: If there is a genuine choice of operator, the relevant Autonomous Community governments can scarcely be denied that choice for their own internal services. This is presumably the main aim of Arriva in Spain: To open up the large, but historically often uncontested, contractual market for public transport. A rather different aim to those market entrants focused on a handful of commercial near-aviation markets, such as Madrid-Barcelona. The hostility of regional government in Spain to Renfe is well documented in Catalunya, but from Aragon to Madrid to Extremadura, none has much love for Spain’s national railway operator. Regional government theoretically counters centralist Spain, representing a powerful ally against Renfe, even if the Catalan example suggests that such power is no guarantee of success. In some regions, Renfe could become vulnerable at much the same time as the local bus concessions that were last extended without contest in the 2000s: In Galicia, law 5/2009 extended most until the end of 2019. The metropolitan concession in A Coruña is already under intense pressure from the CNMC to be modified for compliance with European state aid rules. While all this raises the possibility of multi-modal contracts, especially attractive to an operator with the breadth of Arriva, the more immediate outcome should be greater clarity on the costs of delivering different services: As demonstrated in Catalunya, even apparently “fair” funding mechanisms can disguise substantial cost differences. In a similar vein, the Spanish government’s blanket OSP allocation is eminently challengeable as a policy: Aside from mocking its intended methodology by waving efficiency and utilisation targets for every service that didn’t meet them, INECO’s “rubber stamp” analysis ignores other transport modes, offering no reasoned assessment of Renfe’s contribution to policy objectives such as local mobility: Why support a train between Ferrol and A Coruña with an average of just 16 passengers a trip, when the route is more-or-less mirrored by a faster and more frequent bus?

As explored in The Art of Public Competition sequence of essays, Ferrol-A Coruña does not pose the rational economic question it may seem. The implicit long-run expectation is that the railway service will be improved to compete effectively as a balanced duopoly – although with up to a billion euros of investment required, those improvements may be a long time coming to Ferrol. Liberalisation implies the transformation of the societal model of competition, such as where discrete local duopolies maintain balance, to a more economic model of competition, where factors like cost and demand determine balance across a whole. A Spanish legislature devoid of absolute power can only expose the new model to an environment defined by the old – Spanish policymaking is necessarily structured in a way that lets its society test new legislation, successful EU “directives” included. The challenge for liberalised market entrants is thus to transform the old environment into a new model. Being more operationally efficient or market orientated than the status quo should be straightforward for veterans of open public transport markets. But if the terms of that competition don’t gain state (in the widest sense, of knowing) acceptance, then the theoretically liberalised market might be overwhelmed by attempts, however irrational, to the maintain prior “competitive” balances. As the operator of the Ferrol-A Coruña bus, Arriva presumably have some understanding of the challenge posed by Spanish railway liberalisation, and some strategy for addressing it. The three contexts outlined above – market redefinition, operator contestability, culminating in a regional endgame – give some clues as to why Arriva’s first foray into Spanish passenger railways is on familiar territory for Arriva as a bus operator (both in Galicia and northern Portugal). But is Arriva’s A Coruña-Porto passenger railway service a feasible commercial proposition, or is its purpose more… strategic?

In this essay:

On the Border

European Union directive 2012/34, the “4th Railway Package”, is the guiding policy behind the contemporary liberalisation of railways within member states. Agreed in 2016, active from 2020, the package initially promises open competition on the commercial routes of national railway networks. From 2023 non-commercial public services will become subject to competitive bidding, although prior contracts will continue under 2007 regulations, which allow 10 year contracts with up to 5 years of extension. More protectionist governments can therefore maintain their national railway operator as the sole contractor of subsidised services into the mid-2030s. Spain’s extensive OSP network, with its sub-commercial fares, might therefore be expected to prevent or severely impair competition on local and regional services for some time. However, the fickle politics of contemporary Spain can no longer offer the policy certainties of previous decades: The collapse of Mariano Rajoy’s government will presumably stall the expected June 2018 signing of Renfe’s new 10-year contract. The replacement left-leaning government of Pedro Sanchez has no instinctive alignment with liberalisation policies, but may place greater emphasis on local sustainability, ergo less on national railways. Having been substantially empowered by Catalan and Basque nationalists, it will surely be difficult for Sanchez’s government to promptly sign away another decade of hope for greater autonomic (regional) influence over railways: Exposed to the complex regional politics of Spain, Renfe’s OSP monopoly might erode quicker than expected. Those campaigning outside El Procésfuneral, the ceremonial formation of the new Catalan government that was held within an hour of Sanchez taking office, were demanding civic improvements rather than republics – which may herald a policy shift back towards more practical autonomic issues, such as who runs the trains. Longer term, the rise of Ciudadanos, whose manifesto more clearly favours competition, service and transparency, could yet shift Spanish politics more firmly toward liberalisation.

The 4th Railway Package recognised a potential cabotage loophole in the protection afforded to domestic markets: “(18) The introduction of new, open-access, international passenger services with intermediate stops should not be used to open up the market for domestic passenger services, but should merely focus on stops that are ancillary to the international route.” Specific regulations followed, enacted in the 4th additional provision of Spain’s Ley 38/2015, for which the Spanish competition commission, CNMC, published its criteria in September 2017: 30% of passengers and 20% of distance must be international. It is on this criteria that Arriva have proposed their “La Coruña-Porto” service, effectively under the 3rd “international open access” Railway Package of EU policy, not the 4th. And thus, if approved, Arriva’s proposed operation could (as discussed below, it is unlikely it actually will) preempt 2020’s liberalisation of long distance services wholly within Spain. This context explains why Arriva did not initially inform the Portuguese authorities, and why much of this essay is couched in the context of Spain not Portugal – administratively Arriva’s request stems from Spain.

Spanish railways have been insular since their inception, famously built to a different track gauge to the rest of Western Europe, even though many early routes were funded by French capital and engineered with French expertise. Portuguese and Spanish gauges were eventually standardised at 1668mm, but it was not until the construction of the high speed AVE route from Madrid to Sevilla in 1992, that national railway infrastructure was built to International (“Standard”) gauge – and only with the opening of the Perthus Tunnel in 2013, could International-gauge trains operate internationally without passing through gauge-changers. Gauge differences are the most obvious of a range of technical differences, such as signalling systems, which a future European railway network may eventually overcome. But while the technical standards of Spain’s railways are increasingly European (for example, Spain has extensive plans to implement the ERTMS European signalling and communication standard), the administration of cross-border services remains almost oblivious to common membership of the European Union, let alone the expectations of free movement fostered by Schengen:

For example, traditional restrictions on cabotage continue to apply at the eastern Pyrenean border at Portbou/Cervera – Spanish Renfe trains cannot pick up passengers in France, only set down, while conversely French SNCF trains cannot pick up passengers in Spain, only set down – with the result that each passenger train operated across the border carries passengers in one direction only, and the effective passenger service is reduced to just four journeys per day. A comparable situation on the western frontier was exposed when SNCF upgraded its trains on the TGV Atlantique corridor to use rolling stock without the formal certification to operate into the Spanish border town of Irun, and thus trains that formerly crossed the border to set down passengers in Spain instead terminated at the French border town of Hendaye. Between July 2017 and April 2018 the connecting Sud Express, the overnight train from the French border to Lisbon, continued to depart from Irun, since it traditionally had only been able to set down passengers arriving from Portugal in France, not pick them up on their return, creating an embarrassing gap in the historic Paris-Lisbon route.

Each Iberian cross-border railway service is subject to a bespoke joint agreement between Spanish state operator Renfe, and state operators “Comboios de Portugal” (CP) in Portugal or SNCF in France. For example, Sud Express is operated solely by CP, but rents Renfe (Talgo “Trenhotel”) rolling stock, and runs combined with the overnight “Lusitania” to Madrid between Lisbon and Medina del Campo. While CP entirely subsidises the Sud Express – which primarily links Portugal and France, and is thus of marginal relevance to Spain – the loses of the Lusitania, which links Spain and Portugal, are shared between CP and Renfe. Renfe’s preferred approach, according to former CP president Manuel Queiró, is to form a joint company, “to avoid becoming direct competitors”, which is how Renfe and SNCF operate high speed services across the Pyrenees, the only service between Spain and France that penetrates beyond border towns. Historically Spain’s public transport concessions could not be shared between entities, so joint companies are a culturally acceptable way of managing contested operating environments, if questionable in a theoretically liberalised international market.

The rationale for Europe’s national state railways to support, or even be the slightest bit interested in, international services is rarely considered. Many long distance international services, such as the Sud Express, were introduced commercially by Wagons-Lits, and only when (after World War 2) they ceased to be commercial, did such services default to the (by then) state operators that had hitherto simply provided traction for Wagon-Lits’ private services. Between the rise of budget aviation, faster daytime railway services, poor utilisation of sleeper stock, and post-Crisis financial austerity among state operators, Europe’s international night-train network has gradually evaporated. The EU‘s Trans-European Transport Network policy still perpetuates the outdated mythology of a long-distance European passenger railway network, almost oblivious to its own aviation policy, which means passenger journeys of more than 3 hours will almost always default to air. Meanwhile, in Iberia at least, mid-distance cross-border journeys where rail is a realistic proposition are at best low priorities, neither directed by European macro-strategy, nor of particular interest to nationally-focused railway entities. Surely an opportunity for a liberalised railway market.

The Rise of the Celta

A direct passenger railway service between Porto and Vigo, currently known as the Celta, first operated in 1913. On the Portuguese side of the border the service operates along the 134km “Linha do Minho”, inland from Porto to Nine (the junction for the largest intermediate town in the corridor, Braga) and along the Atlantic coast via Viana do Castelo to Valença, then north across the border, 50km via Tuy (Tui) and Guillarei (the junction with the Vigo-Ourense line) to Vigo (on an indirect trajectory from the north-east, not the south). Journey times between Porto and Vigo improved little before the last decade – over 4 hours in 1980, and still three and a half hours in 2005, much of that time saving apparently attributable to the removal of customs borders under the EEC. Subsequently journey times were improved, primarily by omitting stops at most intermediate stations – the current service stops at just Nine, Viana do Castelo and Valença – reducing duration to two and a half hours. However, overall average speeds of 80km/h within Portugal and 70km/h in Spain, are still slower than the modern autopista (motorway) between the cities (E1/A3/AP-7), which can be driven in about 90 minutes. More local services are offered between Porto and Valença (by CP) and Valença and Vigo (by Renfe) – although the Valença-Vigo service’s timing more-or-less matches that of the Celta service from Valença to Vigo, merely stopping at intermediate stations, and thus the four passenger trains that cross the border in each direction still render only two long-distance international journey opportunities each way: While railway journey times have substantially decreased, Celta frequencies remain (with the exception of a short-lived additional service in the mid-1980s) unchanged, with one morning and one evening train in each direction. With no substantial change in overall service pattern over those four decades, two separate trains (sets of rolling stock) are still required to deliver what is now a total of just 10 train-hours of operation each day. Such history is often indicative of a staff-centric operating environment where working patterns are firmly rooted, perhaps inhibiting the commercial-style business planning that might here seek to use the resources freed by faster journey times, to either reduce operating costs or increase service frequency.

The Porto-Vigo service is currently operated by CP, but uses rolling stock leased to CP by Renfe for 315,000 euros per train annually (2010 prices), and is crewed by staff from both operators – for example, the driver involved in the O Porriño derailment was reported Portuguese, while the ticket inspector was Spanish. The service’s current rolling stock was built in the early 1980s and is now reaching the end of its working life, but represented a modernisation from prior CP stock that was designed in the 1960s. Since agreed at the 2013 XXVI Hispano-Lusa summit, tickets on cross-border services are sold by Renfe, at a fare just below the combined price of the local fares on each side of the border. This “international” fare structure may be pragmatic given the ease of buying a second ticket at the border, but since both local fares reflect their subsidised domestic environments (currently €11.10 Porto-Valença and €4.00 Vigo-Valença – under 10 cents per kilometre), the international operation is particularly hard to extract from the surrounding state-funded public service obligations. INECO’s 2017 review added two daily local Tuy-Vigo trains, which currently start at Valença, to Renfe’s OSP, so Renfe will retain the contractual flexibility not to offer a cross-border journey at a publicly subsidised price. However, it is scarcely practical to simultaneously offer a fully commercial fare from Valença to Vigo and a substantially subsidised fare from Tuy to Vigo, when Valença and Tuy are separated by nothing more than a bridge across the river Minho. So even if the international Porto-Vigo Celta service were operated without any public support whatsoever, the service’s fare (and thus revenue) potential would still be framed by the public support of more local services. Such an environment surely dissuades operators from increasing frequency on what remains quite a locally-focused service.

Much of the current Celta service stems from threats to abandon the cross-border passenger service entirely. Herein a distinctly Iberian tale of two extremes:

  1. The ambitious, but not necessarily cost-effective, proposal to build a high speed AVE route between Vigo to Porto, which would link the anticipated AVE networks in Galicia and Portugal and allow rapid travel along the Atlantic coastal axis from A Coruña to Lisbon. Since the project did not accurately fit into the EU’s (Trans-European Transport Network) Atlantic Corridor, which primarily aimed to link Iberia and Northern Germany, funding was especially challenging, and by 2005 the project was already stalling. In principal the proposal was agreed again in November 2009, but shortly thereafter the financial impact of the 2008 Crisis curtailed all Portuguese rail development.
  2. The operational reality of the existing service between Porto and Vigo, international patronage having declined to near-insignificant volumes for a railway service: Between 2001 and 2004 almost 8,000 passengers were lost, with just 15,500 annual passengers remaining in 2010, a reported average of just 11 paying passengers per train. CP had first proposed abandoning the cross-border service in 2005, but by 2011 their intention was clearer – indeed, other territorially marginal CP services, such as the Elvas-Badajoz border crossing, did cease in 2011. CP’s proposal would have only affected cross-border services, not Porto-Valença, and thus only disadvantaged international passengers and those in the Spanish border town of Tuy, who were keen not to become such a footnote.

The first was primarily a Spanish perspective, the second primarily Portuguese – and almost simultaneously in 2011, Spanish interests, including the regional Galician government Xunta, were demanding progress on the Vigo-Porto AVE, while the Portuguese economist and minister Álvaro Santos Pereira was declaring the whole scheme “financial madness” and seeking ways to reduce CP‘s existing costs. During 2012 a shared perspective started to emerge, Renfe accepting AVE could not be the solution to everything, CP encouraging Renfe to shoulder some of the cost of providing the cross-border element of the existing service. From this shared reality emerged more practical plans to enhance the service’s most obvious deficiencies, including viable through-ticketing, accelerated journey times by removing minor station stops, and marketing the service as the “Celta”. Annual patronage rose steadily from 23,500 in the first year of service improvements (2013), to 80,200 in 2016 and 91,600 passengers in 2017 – a more respectable average of 63 per train, albeit only yielding annual revenue in the order of a million euros – covering operational staff and infrastructure charges, perhaps even train rental, but not yielding sufficient surplus to yet justify service enhancements.

The government of António Costa heralded renewed interest in Portugal’s railway modernisation: “Ferrovia 2020” focused specifically on improving existing lines, notably the electrification of the Linha do Minho all the way to the border, ahead of the historically favoured priority of Madrid-Lisbon – a subtle, but none-the-less significant policy shift away from the priorities of centralist Spain, toward the Atlantic Axis, even if it did not meet Galician expectations of an AVE. Complementary work by Spain’s ADIF between Vigo and Tuy should allow the use of electric trains on the Porto-Vigo route in 2019, with journey time reduced to 90 minutes. Desires for a Vigo-Porto AVE still linger in Galicia, currently placated by Spanish plans to open a southern route into the new “AVE” station of Vigo Urzáiz, allowing services from Porto a more direct access to both the city and the future Galician AVE network. Meanwhile CP plans to buy a fleet of 28 trains, 6 specified for long-distance, at least some bound for the Celta, potentially allowing a significant increase in frequency. Speculation that CP’s new rolling stock could be hybrid and gauge-changeable (thus able to travel throughout Galicia, but at likely half capital cost of Renfe’s hybrid multi-gauge AVE stock – which, at €21 million a set, is surely too valuable to be gambling on a cross-border route with the performance history of the Celta), adds further weight to prior reports that CP and Renfe intend to form a joint company to exploit the A Coruña-Lisbon route after 2019. And it is in this environment that Arriva now proposes much the same.

¿Arriba Arriva?

Much commentary has centred around the lack of practicality of Arriva’s formal service notification. Their proposed start date, 17 July 2019, is itself curious – mid-summer, and not the standard (European) second Saturday of December. Arriva’s proposal presumes much of the infrastructure improvement described in the last paragraph, including 25 KV electrification throughout, which even Arriva acknowledge is likely an aspiration for 2020. Traction is consequently not specified, only the use of two sets of 220-seat “Talgo Pendular” coaches, to be maintained at Redondela (near Vigo) – and even those are subject to a “pre-operational” study. The curtailment of many overnight trains has created a glut of theoretically convertible coaches, so finding two spare sets of Talgo coaches seems plausible. Candidates for traction are primarily limited by a requirement for 200 km/h running in Galicia – logically provided by Renfe’s class 252 locomotives, four of which are already maintained at Redondela, but unfortunately a class with no prior tradition of non-Renfe operation which is held by an organisation with no intention of renting its rolling stock – even if the CNMC doth protest. The default “Iberian open access” locomotive, Stadler’s Euro 4000, is already operated by Takargo for freight from A Coruña onto the Linha do Minho, but is not electric and critically cannot reach 200 km/h, which only Renfe’s class 334 variant can. Entirely new trains are infeasible within Arriva’s intended timescale – even if new Iberian-gauged stock could be ordered and delivered in a year, an extensive period of safety-related homologation would inevitably delay the carriage of passengers. Traction may also transpire to be legally significant, because Spanish legislation frames a “railway company” as strictly that which provides the traction.

Track-work on the Spanish side of the border is assumed by Arriva to allow at least 120 km/h operation from Vigo to Porto, within a 140 km/h restriction on Talgo stock in Portugal, which helps explain why the service does not terminate in Lisbon, something Arriva acknowledge as an aspiration within their proposal: The line from Porto to Lisbon is operated at up to 220 km/h. Arriva’s timetable, which proposes departures every 4 hours from 08:00 until 20:00 from A Coruña, returning 07:00 (local time – offset one hour from Spain) and every 4 hours from Porto, is as conceptual as the rest of the operation, taking no obvious account of existing train paths of single-track sections, only that sufficient unused train paths exist. Finally nothing is said of how the trains will be staffed, which may yet be the biggest single challenge to the fledgling railway operator. Portugal remains the most obvious source of staff, where commercially trained drivers are reported to be lacking employment. Overall, while Arriva’s proposal is easily dismissed by a railway culture expecting precision and finding none, as a broad initial concept the proposal shows intent, which is perhaps the most that can reasonably be expected at this stage of railway liberalisation.

Arriva’s proposal represents a clear improvement in route, time, frequency and vehicle quality over the status quo, but, as described in the previous section, the route was already likely to attract such service improvements. Doing what you competitor might do, just sooner, is a legitimate form of competition, although finite track capacity adds a complication: While the two hundred unused daily paths between Vigo and Santiago de Compostela are unlikely to be overwhelmed by sheer volume of competing services, track capacity further south, specifically on the single track section between Nine (the junction for Braga) and Valença (the Portuguese border town), could fill quickly. Ultimately such demand for train paths might spur on further track upgrades – a model of demand-driven infrastructure development demonstrated by DB/Arriva’s Chiltern Railway – but in the short term, securing train paths by occupying them first is a pragmatic approach: Officially, Infraestruturas de Portugal (IP, the Portuguese track manager) prioritises based on “public use, particularly services carried out under a public concession contract”, which will favour existing CP services, but thereafter should favour an actively used Arriva international service over any subsequent proposal. Granting a commercial operator track capacity that might subsequently limit the ambition of a state operator would be a first real test of the supposed organisational separation between track and trains.

Overall, Arriva’s proposal should test the willingness and reaction of “the state” on a range of liberalisation topics – from an expected reluctance to yield assets, to a competitive structure that disrupts traditional planning processes. As discussed in the introduction, Arriva’s strategy is contextualised by far more than just head-to-head (albeit here pre-emptive) competition. Two themes emerge strongly from the text of their proposal:

  1. The close territorial relationship between Galicia and northern Portugal – “dinamismo socioeconómico” and “intensas relaciones de movilidad”.
  2. The scope for product differentiation by integration of rail and bus, both local and interurban.

While both are ultimately within the bequest of the state(s), it can be argued that both are easiest for Arriva to achieve: Multinational businesses can cross borders locally, without having to work through cumbersome sovereign structures, while Arriva is uniquely well placed to sell tickets across the many local networks it operates, even if its commercial freedom in Iberia is not unlimited. All this is important, because, as discussed in the introduction, the first stage of liberalisation is to build a societally acceptable competitive model – to offer that which “the state” struggles to achieve – which for Arriva is a combination of policy (regional cohesion along the Atlantic Axis) and service (public transport inter-modality): Not just competing against existing providers for their existing business. Such caveats are commonly asserted by businesses that are privately intent on simply competing against existing providers for their existing business, but here creating a “new demand” genuinely matters for the long-term success of liberalisation, and that makes Arriva’s approach interesting.

The balance between new and existing demand is naturally difficult to assess. In anticipating a minimum of 0.36 million passengers per year, with the potential for 0.8 million, Arriva’s aspirations clearly exceed the roughly 0.1 million of the existing Celta service, but are relatively modest in comparison to the 11.2 million they cite for the “Galicia-Lisbon” corridor, or even the 10.4% of that they attribute to existing scheduled public transport. Approximately one million of this “Galicia-Lisbon” market consists of overland journeys to or from Porto airport, a market that will remain best served by direct cross-border bus services (because of the lack of direct railway access). Even with 90 minute rail journeys between Porto and Vigo, existing bus services will remain somewhat competitive between the two city centres – offering more services (10-14 daily in each direction, albeit spread across 3 competing operators – Autna, ALSA and Flixbus), with commercial fares typically at or just below the existing rail fare, creating little scope for competition on price. Rail’s most obvious advantage lay in places north of Vigo (Pontevedra, Santiago de Compostela and A Coruña) where track speed is greater than road, and direct bus frequencies are lower – for example, both ALSA and Flixbus take 5 hours to reach A Coruña from Porto, a journey Arriva propose to complete within three.

275 Days Later

One particularly curious number stands out in Arriva’s proposal – two hundred and seventy five. 275 is the number of days per year Arriva’s service would operate, which the proposal notes would be distributed according to demand and seasonality. 275 days implies more than a simple work-day (weekdays except public holidays) service. The closest approximation is a pattern more common in Spanish urban bus operation – Monday-Saturday, except public holidays and August – work-related demand typically halving in August due to holidays. Plausible given the intended integration with bus services, although it supposes a primarily business, rather than leisure market. Autna’s Vigo-Porto schedule offers 5 services on weekdays and only 3 at weekends, confirming said bias to business travel, but ALSA’s service level would appear to increase toward the weekend, suggesting the opposite. Even if Arriva were proposing a service better tailored to business travel, it would be curious not to operate on public holidays, which can be notoriously busy days for long-distance travel, so perhaps “275” equates to weekdays plus high-demand public holidays? Perhaps the intention is even more subtle, such as operating only one of the two train sets on certain days, possible within the terminology of 275 “Equivalente Días / año”. Whole days with no active service would allow longer continuous periods (beyond nights) for rolling stock maintenance, thereby requiring only two sets of rolling stock, although serious technical failures could still be crippling.

Eight daily departures over 275 days, is 2200 per year, multiplied by 220 seats, is a total of 484,000. The minimum anticipated patronage (0.36 million) would imply roughly 75% occupancy, while the potential patronage (0.8 million) equates to 165%. An overall occupancy of 75% is possible, if challenging – Renfe achieved it on AVE with heavy price-based promotion, although they only matched the 90% typical of budget aviation on the busiest routes, from Madrid to Barcelona and Sevilla. Since long distance services cannot happily exceed 100% occupancy at any one time, if Arriva’s service is to obtain its 0.8 million annual passenger potential – 165% occupancy – a substantial proportion of seats will be used more than once: Every seat used twice implies at least one of those uses is for an entirely domestic journey. The CNMC only requires 30% of passengers are international, and Arriva even acknowledge in their proposal that domestic patronage would be required to make the project “economically viable”, but these derived occupancy rates give a first indication of how dependent the service might be on domestic patronage: Assuming the potential patronage of 0.8 million a year, and an optimistic 80-90% of available capacity sold, roughly half of journeys would need to be entirely domestic.

Unless Arriva’s integration plan develops some hitherto hidden travel pattern, there will be minimal demand for Arriva’s railway service wholly within Portugal: Porto-Valença is the only tangible demand served, yet Valença’s 14 thousand residents barely feature in the crudest of demand models for the service, that based on the pattern of population: The cities of Porto, Vigo and A Coruña dominate, each populated in the order of 200 thousand people, the population of the entire urban area of Porto a far larger 2.9 million (but a part of that is too remote from the railway line to be directly comparable), with approximately another 200 thousand population served by the stops between Vigo and A Coruña (Pontevedra, Vilagarcía de Arousa, Santiago de Compostela). Ergo almost every domestic journey will be carried within Spain, and thus along the Vigo-A Coruña corridor. Assuming revenue is maximised by carrying passengers furthest, and rather crudely assuming those three roughly equal population blocks within Galicia (Vigo, A Coruña, and Pontevedra/Vilagarcía/Santiago) have equal pull toward Porto, each of those three blocks loads 33% of the train’s capacity to/from Porto, which means 33% is available Vigo-A Coruña, and a further 33% is available by Santiago de Compostela for the final stage to A Coruña. These assumptions yield an optimised load balance, filling every seat on every train between every stop, that almost exactly equates to 165% occupancy overall. Given 220 seats per train, seats would be allocated in batches of approximately 73 to each of these origin-destination pairs:

  • Porto-Vigo
  • Porto-Pontevedra/Vilagarcía/Santiago
  • Porto-A Coruña
  • Vigo-Pontevedra/Vilagarcía/Santiago/A Coruña
  • Pontevedra/Vilagarcía/Santiago-A Coruña

This broad pattern suggests 60% of passengers would be international, the remaining 40% carried domestically on the Vigo-A Coruña corridor. Such crude analysis implies Arriva’s service would, first and foremost, be international between Galicia and Porto, and not merely a ploy to undermine Renfe domestically within Galicia. Ticket sales could theoretically still be concentrated within the Vigo-A Coruña corridor, but that would imply loading a separate set of passengers specifically between Vigo and Porto, fundamentally skewing the underlying (population-derived) market, which in turn implies heavy discounting of fares on the Vigo-Porto leg. Not only might such discounts offset any revenue gains from focusing on Vigo-A Coruña, but, with three bus competitors between Vigo and Porto, such discounting could trigger a price war – which given the higher operating cost of rail, Arriva would not expect to win.

The cynics are correct in so far as the Vigo-A Coruña corridor could be attractive to new railway operators, not least because the underlying market is almost perfectly distributed within the corridor, allowing optimisation-centric operators to fill every seat between every stop, without recourse to complex pricing strategies. Renfe currently offers 9 or 10 daily fast services in each direction between Vigo (Urzaiz) and A Coruña, and another 7 or 8 slower stopping services (from Vigo Guixar). Additional Ourense corridor trains serve Santiago-A Coruña, but these are of only marginal relevance to the Arriva proposal. Unsurprisingly, both existing direct services perform rather well, amongst the best-loading regional railway services in Spain:

  • The fast service conveys 1.243 million passengers per year (2015 data), an average of 249 occupants per train, 89% occupancy.
  • The slow service conveys 1.166 million passengers per year, an average of 169 occupants per train, 85% occupancy.

Couching the Cost

Renfe’s combined Vigo-A Coruña service earned €14 million in 2015, an average fare of €6. That value is broadly consistent with the route’s fare structure, after adjusting for the quarter of journeys that are multi-trip or similarly discounted. Operating the associated 1.7 million train kilometres cost Renfe €25 million. The 2012 benchmarking of European railway operators did not include Spain, did include capital investment (which Renfe typically excludes), and excluded track access charges (which Renfe includes) – so is not directly comparable to Renfe’s costs, but will give some indication of Arriva’s costings. The European average of £11.33 per regional train km, roughly €13, suggests a cost of around 22 million euros for the current Vigo-A Coruña corridor service, before infrastructure charges of 2-3 million: Close enough to existing revenues to make the Vigo-A Coruña corridor potentially vulnerable to commercial competition – especially at the margins, but potentially also just by filling every seat between every stop. Vulnerable unless fickle track access regimes shift to be significantly less favourable to operators – a potential strategy for a protectionist state that would instead subsidise the public service operator more directly.

In a proposal full of variety, another: Arriva’s service will pass over the tracks of all three Iberian national railway infrastructure managers – ADIF, ADIF-AV, and IP – each of which document charges in their network statements. For this analysis, electric power is costed even where there currently is no electrification, fees based on comparable electrified routes. ADIF-AV‘s station passenger tariff assumes Arriva’s potential patronage, distributed based on the five optimised origin-destination pairs derived above – with Arriva’s minimum patronage total infrastructure costs would be around 7% lower. ADIF‘s A Coruña stop cost averages separate origin and destination charges. Train parking/stabling costs between services are not included, but only likely to add a few percent to total infrastructure costs. The table below summarises Arriva’s infrastructure access costs per one-way train journey. In Spain charges are primarily for stations, while in Portugal charges are almost entirely for track, although both ADIF and IP ultimately levy a net similar €1.5/km (ADIF-AV’s additional €0.41 tariff per passenger boarding/alighting an international train, inflates their charge significantly). Non-infrastructure costs dominate: Assuming train operating costs of €13/km, the 342 kilometres between A Coruña and Porto total about €4450, to which infrastructure adds just €580. The total cost per Arriva Celta train journey would be about five thousand euros, about 11 million euros annually to operate the entire proposed service.

Infrastructure Access Costs per Arriva Celta Train Journey
Network Track Stations Total
ADIF €58 €112 €169
ADIF-AV €49 €155 €205
IP €208 €1 €209
Total €314 €269 €583

Both the existing Renfe Vigo-A Coruña corridor services are OSP, a decision which partly reflects an operational analysis of the ratio of seat-distance to passenger-distance, where high overall occupancy still equates to over half the seat-kilometres operated being empty – overall 85-90% occupany disguising a high proportion of seats reused over the course of the train’s journey – many relatively short passenger journeys. This must be placed in the context of a traditional fare structure where passengers pay primarily per kilometre. As shown in the table below, the rate per kilometre for adult single fares from Vigo declines only slightly with distance – Vigo-A Coruña fares obtaining a per kilometre rate only about 10% cheaper than the far shorter Vigo-Pontevedra. In contrast, commercial fare structures, as already practiced on AVE, taper more with distance, and thus short-distance passengers tend to generate more revenue per kilometre. Thus well-balanced routes like Vigo-A Coruña perform much better under commercial logic, than under the OSP logic that concludes state support. Vigo-A Coruña is an intriguing example of a route where both commercial and state logics might validate their own criteria, and thus both commercial and state actors may expect to operate the service themselves. Yet again, Arriva’s proposal serves to test the new against the old.

Adult Single Fares from Vigo by Distance, June 2018
Vigo To KM Fast Train Slow Train
Fare Fare/KM Fare Fare/KM
Pontevedra 28 €3.65 €0.130 €3.20 €0.114
Vilagarcía de Arousa 54 €7.35 €0.136 €6.10 €0.113
Santiago de Compostela 95 €11.10 €0.117 €9.25 €0.097
A Coruña 156 €18.10 €0.116 €15.30 €0.098

Portugal’s fare structure fits much the same pattern, with slightly cheaper fares on slower services, and only a slight reduction in the rate per kilometre with distance. Fares on the Linha do Minho are, however, 20-25% cheaper per kilometre than the Vigo-A Coruña corridor, for example the 134km from Porto to Valença is priced by CP at €11.10 fast and €10.65 slow, while Galician prices for the same distance would be approximately €15 and €13 respectively. Renfe’s €4.00 for the 52km Valença-Vigo may reflect Portuguese fare-scales, or may be a pragmatic reflection on the fact the direct (road) distance is much shorter than the rail. Regardless, as discussed earlier, the full Celta fare between Porto and Vigo is basically just the addition of each local fare – €14.95 or €23.90 return – which makes the current fares on the southern part of Arriva’s proposed route 20-25% cheaper per kilometre than the northern part. The quid pro quo, that the northern part is faster, and thus entirely reasonable to a user-base that expects to pay more to go faster.

A Liberalising Proposition

The baseline assumption is that current fares are simply mirrored by Arriva. The resulting revenue per one-way train journey, for each of the five 73-seat journey legs derived earlier, is shown in the table below. The average fare is the single adult fast fare, reduced a third to account for discounts such as returns and concessions. With every seat sold, the maximum baseline revenue per train-journey would be around €4500, a total of 10 million euros annually. Assuming the same balanced distribution of passengers, Arriva’s minimum expected annual patronage (0.36 million) would yield 4.5 million euros.

Baseline Maximum Revenue per Arriva Celta Train Journey
Journey Average KM Average Fare 73-Seat Revenue
Porto-Vigo 186 €10 €730
Porto-Pontevedra/Vilagarcía/Santiago 245 €15 €1100
Porto-A Coruña 342 €22 €1610
Vigo-Pontevedra/Vilagarcía/Santiago/A Coruña 59 €7 €510
Pontevedra/Vilagarcía/Santiago-A Coruña 97 €8 €580

So even if Arriva’s A Coruña-Porto service were to reach its maximum potential patronage, if revenue was only derived from existing fare structures their service would be at best a marginal commercial proposition – not quite covering the full costs estimated in the previous section, of €11 million/year, but perhaps covering its day-to-day operating costs (without 10-20% for capital expenditure). Those costs disguise many unknowns, even to Arriva, but are unlikely to be so inaccurate as to alter that assessment of a likely lack of profitability. A strictly rational analysis would conclude Arriva planned some alternative fare structure, yielding greater revenue per passenger than existing operators, without provoking direct competition. There are clues to this littered throughout their service proposal:

  • The dominance of existing fare structures wanes when connecting locations that would otherwise require a complex sequence of different services and fares. Ferrol-Porto is the perfect example, a pairing that currently defies automated comparison, which Arriva would be uniquely placed to sell tickets for. Even Ferrol-Vigo, for which Renfe can offer no sensible journey. Ideally integration should open genuinely new markets. Integration that simply offers through tickets to otherwise existing customers is risky because it can lose revenue in a manner that is difficult for a multi-business operating group to account for.
  • The 275 days of annual operation suggests targeting, perhaps to busy periods when competitors’ capacity is filled, but more obviously to a wealthier market – specifically people that hitherto rejected the train between Vigo and Porto because of service or vehicle deficiencies. The Talgo rolling stock proposed by Arriva is not a low-budget option, and while Arriva’s service may only have half the speed of AVE, the passenger environment could match or exceed that of AVE stock. Market this “souped-up” Celta such that Atlantic Axis travellers become convinced this is, in all but name, the AVE they’ve been waiting for, and a vastly more profitable fare structure opens up. What’s unclear is whether AVE justifies its higher fares specifically because it goes faster, reflecting the cultural norm seen in traditional fare structures, or because of AVE’s brand image or perceived national role, which imply marketing alone could be successful.

A business plan of sorts, but lacking certainty. For a transport operating group with over 40 billion euros of revenue in 2017, a potentially loss-making service on the scale of Arriva Celta may be an acceptable risk, if it achieves something else in the longer term. As discussed in the introduction, even if Arriva were to open up the railway passenger market between Galicia and Portugal, without breaking the dominance of the state operators, Arriva could not thrive: The failure of a niche service such as A Coruña-Porto would be almost irrelevant to Arriva. But if, when prize contracts such as Madrid’s suburban Cercanías are eventually tendered the only feasible bidder is still Renfe, Arriva would have lost a significant business opportunity. Arriva don’t need an immediate solution because any contest could be a decade or more away. But in the meantime they need to be probing, experimenting, and generally trying to transform the old “competitive” environment into a new model that accepts their business. That’s the true challenge of Spanish passenger railway liberalisation – indeed, liberalisation depends upon it.

In 1992, at the very beginning of British railway liberalisation, Stagecoach Group launched a railway service by attaching a couple of regular passenger carriages to the back of an overnight Sleeper train between London and Scotland, in an attempt to sell the low-price seat-only overnight travel that the (then) state railway operator had discontinued. It was a market broadly familiar to Stagecoach from their long-distance coach operations in the 1980s, but the railway environment was unfamiliar and in the process of change. The scale small, the risk manageable, and ultimately it was a commercial failure for Stagecoach, especially as British railway “privatisation” embraced a franchise model that left little space for commercially speculative services. However Stagecoach’s early experience of rail surely helped the group win the very first railway franchise in Britain, South West Trains, which they held for the next twenty years.

Arriva’s A Coruña-Porto proposal is not in itself innovative, or even unexpected – the incumbant operators were already lumbering toward something rather similar. Much like Stagecoach’s 1992 overnight train, the Arriva Celta should be considered an initial test of a new, potentially evolving regime, in a market territory that is already familiar to Arriva. Arriva’s proposal encompasses a wide range of such tests, from asset liberalisation and basic operation, through product and pricing, to development strategies and regional politics. Arriva’s traditional strength is surely its adaptability to different “competitive” environments. So while Arriva perhaps lacks the commercial flare of the nascent Stagecoach – maybe look to the Asturias for that, for surely none trains drivers there expecting to compete on an AVE line – Arriva’s traits may make the group’s Iberian railway strategy key to the success of liberalisation in the long term: A peninsular where the scope for purely commercial competition is rather limited, and many passenger railway services will inevitably remain in the care of a state that has, at least in parts of Spain, traditionally maintained its own competitive model.

Update: On 30 April 2019, the CNMC finally declared that Arriva’s Celta proposal would not have a negative impact on the economic balance of Renfe’s OSP operations.

The title illustration is intended as metaphor, which may first require description: Talgo coaches being shunted out of their Sant Andreu (Barcelona) depot, in search of a proper mainline locomotive, which is surely waiting on the other side of the endless AVE construction site that might, one day, open as Sagrera station.