This essay analyses and explores the regional passenger fare structure of Renfe, Spain’s national railway operator. The question, “Is Alta Velocidad Fast?”, derives from Renfe’s tradition of pricing slower trains cheaper. The question asks whether, in the era of yield management (balancing current patronage to current capacity by modifying price), the traditional fare structure should be applied to high speed, AV, operations? The journey provides an insight into the structure of modern transport geography, the haphazard strategic development and exploitation of Alta Velocidad, the management of national inequalities through fares, the conflation of public and commercial roles within single shared operations, and, from a perspective other than infrastructure, the contemporary challenges to Spain’s railways.
The introductory section describes the current trend toward Temporal Ticketing, with a reflection on how this alters transport geography and state: While the growing reliance on algorithms challenges established operator dominance, it is not necessarily incompatible with societal behaviour, especially for less familiar journeys. A brief history of Avant follows, the Regional Alta Velocidad funded as a public service obligation (OSP), with extensive analysis of fares on Avant corridors, including an assessment of the selective use of yield management on parallel commercial AV services. Avant is placed in context by a similarly detailed analysis of fares on slower regional services, those on which passengers are Paying to Go Faster. The strong relationship between fare and speed is confirmed, but with a balanced structure of revenue-to-cost that equalises quite different styles of operation.
Fare’s Fair exposes the differences in fares and speeds between Spain’s regional Autonomous Communities, demonstrating how fares have apparently been used to manage the inequalities between regions. Unfortunately the same structure has not been adequately deployed to counter AV‘s inequalities, ultimately because AV pricing has been defined by commercial aviation markets: By matching, not taxing, aviation, the cascade effects (of inflated expectation) to lesser classes of transport can only be addressed through subsidy – the flawed policy that is exposed by AV Cercanías, a theoretical high speed suburban railway – Avant being the closest non-theoretical product to AV Cercanías. Inter-City probes how Renfe has evolved the management of marginal longer-distance trains, especially over non-AV routes, where the same train may share regional public service and inter-regional commercial roles. The conflation of pricing models is highlighted, and evidence is presented that suggests the state is subsidising the train, and not specifically the seats upon it – a pattern that might concern the European courts, were it to continue after 2020.
The concluding section takes a contemporary journey From Extremadura to Catalunya Nord – the origin exemplifying the political impact of AV‘s inequalities, the destination actively challenging the idea of region as strictly administrative. Along the way, how transport geography relates to the source of its finance, how the preoccupation of the Spanish political state with infrastructure inhibits behavioural policy interventions, and ultimately, how Renfe transpires to be a better manager of state than state. The Postscript provides an intense reflection on an otherwise somewhat long and technical analysis of what may seem quite a trivial topic, but actually explains much about the state of contemporary Spain.
The curious case of railway ticketing is one that invariably embeds and perpetuates cultural biases to the perennial question, what is value? For example, the pre-privatisation British Rail ingrained its primitive yield management strategy, that return fares were “cheaper after 09:30”, so deeply into the state psyche that decades later concessionary fares policy was still formulated to protect “the morning peak”, even though the late afternoon had long since become the busier period for many regional public transport networks in Britain. Britain’s railways are still stuggling with the issue, although Britain seems unlikely to shift away from a model that emphasises the temporal capacity of the transport system to handle the intending passenger, and prices journeys accordingly. In parts of Europe, not least Spain, the temporal capacity of the transport system is traditionally not a consideration in pricing. Indeed, every “salida”, the busy days for longer-distance travel at the start and end of holiday periods, the mere possibility that the capacity of Spain’s transport system might have been temporally exceeded, can be rendered a national preoccupation – the media routinely broadcasting reassuring images of half-empty roads and passengers boarding trains like normal. The underlying presumption is that the system accommodates the passenger, even if that means some time periods are far busier than others, and some operational assets lay idle during quieter periods. For example, Barcelona’s metro currently has no fare variation by time of day, even though its patronage is far from constant across the day. The idea of introducing such a variation has been considered only recently, as a means of managing peak overcrowding, and that was only triggered by the post-Independència financial realisation that increasing demand could no longer be met with extra capacity. For state-provided systems, the implicit presumption is that the state will provide. The reasoning is indubitable: State, in its broadest sense, is perpetual, as stable a constant as can be reasoned – not a facet that alters at 8 o’clock just because other people also want to travel then.
Where transport systems are liberalised, most obviously in European aviation, yield management has become the accepted norm. The price of a flight might double or halve from day to day, as airline operators try to fill every seat in every plane – managing their supply, which is largely fixed in the short run, by altering price. Likewise, few will be surprised to discover a flight from Madrid to Berlin (2300km) that today just happens to be cheaper than a flight from Madrid to Barcelona (600km) – in stark contrast to the state-centric model, where fares typically retain a strong correlation to distance. Just two examples of how liberalised transport systems are shifting geography from the spatial to the temporal, and, perhaps more alarmingly, de-humanising geography: Not only by emphasising the economy of the carriage over that of the passenger, but also by encouraging travel specifically when others are not, in opposition to many natural group dynamics. It is in the midst of this tumult that we find the contemporary “Alta Velocidad”, Spain’s high speed passenger railway – a state-implemented network operating in an increasingly liberalised market.
The question posed by this essay, “Is Alta Velocidad Fast?”, is not just the question it may first seem, that which begets the retort, compared to what? Traditional Iberian railway fare structures differentiate between trains based on their speed, where those that complete their journey faster cost more to travel upon. Thus, to quote Franklin, time is money – but here the value of time is constant over time, not varying within time (from hour-to-hour or day-to-day) as is more common in Anglo-Americanised markets. AVE, and similar long distance inter-regional public transport services, are forced by their increasingly competitive liberalised environments to adopt fare structures which undermine the base understanding of state: The national in Spain thereby becomes the liberalised global. Yet within Spain’s regions the traditional fare structure still predominates, and thus perhaps regions better retain their state. Spanish railway liberalisation implicitly acknowledges this by opening up most inter-regional services to competition in 2020, while likely retaining state control (via Public Service Obligation) of services within regions. An analysis that might confer the dominance of regional politics over national, but more specifically, that regions are more knowable among their own populations than the nation of “Spain”, and thus regions evoke greater protection of their state. The strategic counter-balance, that the national will, by dint of improved asset utilisation and reduced reliance on new assets to deal with insufficient capacity, become relatively more efficient economically. However as this essay explores, while local juxtaposes to national, regional is not necessarily aligned to local.
Optimisation for knowing differs from optimisation for analytic efficiency – and herein lay much of the tension between contemporary society and computational optimisation: Since (and arguably before) Euler tried to cross the 7 bridges of Königsberg, transport has been posing computationally challenging problems, from network route planning to supply chain logistics. But it is perhaps only in the last two decades that the average traveller has become directly exposed to such computational optimisations. In the mid-1990s the apocryphal tale was told of an Operational Researcher who commuted each day to his workplace in London Transport’s headquarters near St James’s Park, central London. Upon arrival in the capital at Paddington railway station he was faced with two potentially viable Underground (metro) routes – direct on the Circle line, or via Bakerloo and Victoria lines. For the casual user, the direct Circle line would be the obvious choice, both requiring no change of trains and, based on the stylised London Underground network map, of similar distance. However those distances are actually a lot less similar than the stylised map betrays, and the Circle line had a far lower effective frequency than the alternative route via Bakerloo and Victoria lines. The calculation of optimal route, at the precise moment our expert commuter arrived at Paddington, originally would have relied on train service predictions, but could now be improved with live-time data delivered to a mobile phone application. The optimisation rarely saved more than a few minutes. In the 1990s these were games played by mathematically-minded transportation planners. Now they are becoming the norm among regular travellers, those whose behaviour is optimised by their “smart” mobile phones. The domination of this technology over its human users is clearest among those that happily walk or drive round in a circle, simply because the underlying dataset is missing a network node or link – and thus cannot calculate the direct path that should be obvious to the user, had they not placed absolute trust in their (sadly fallible) device. In the final analysis, these users have become so optimised that they have ceased to know.
This trend is not new: In the spirit of Kuhn, the Enlightenment West has progressively expanded the complexity of its worldview by simplifying every-thing within that world. And given such apparent acceptance of technology over human, yield management of the most familiar of journeys should now seem entirely reasonable: Blind user trust in journey-planning algorithms readily extends to ticketing. In practice the algorithms taking users’ fare decisions would compete with the operator algorithms setting prices, a computer-vs-computer model already successful (except when dealing with the unexpected) in financial market trading. The economic efficiency implied would liberate humans to obsess about something else – and ultimately the minutiae currently associated with transport ticketing would be replaced by trust that the system tends to offer “best value”, with any user interaction reduced to broad concepts such as brand. Transport operators traditionally consider themselves to be the sales channel of their product, and even in the most entrepreneurial arenas (Ryanair vs Skyscanner in 2007) operators are naturally hostile to any optimisation they don’t control. Aside from exerting ownership over data (a battle largely lost), the operators’ only logical defense is to simplify their ticketing structure – logically opposing the yield management that is inextricably driven by market competition. Thus operators in open transport markets will ultimately be reduced to supplying a service sold by technology platform providers (such as Google) or travel metasearchers (such as KAYAK) – a keenly fought technology market that has not yet peaked or consolidated. And once again, in the midst of this tumult that we find the contemporary AVE, its state-owned operator, Renfe, accustomed to a national prominence that it will not yield easily – even if, as discussed in Arriva Celta, Renfe’s role as the provider of national “presence”, beyond mere journey utility, is easily taken by state track owner ADIF, leaving Renfe vulnerable.
Regional Alta Velocidad
The contemporary character of Spain’s high speed railway network, Línea de Alta Velocidad (LAV), was defined by the most unlikely of places: Ciudad Real, the “royal city” of just 70 thousand people, 170 kilometres south of Madrid. LAV, and its high speed trains popularly known as AVE (Alta Velocidad Española), were conceived in the 1980s to serve aviation markets, and were originally intended only to link the largest cities of Spain. Construction of the first line between Madrid and Sevilla (both cities of over a million people) thus commenced with just one intermediate stop planned at Córdoba (population 325,000). Ciudad Real’s station was unplanned, the result of a last-minute political negotiation by the city’s mayor. The legacy for LAV is a network on which less than half the stations handle a million or more passengers a year (boarders and alighters combined), and perhaps for Ciudad Real, an inability to know when to stop building new passenger transport terminals. In the rush to open the line for the 1992 Sevilla Expo, many decisions were taken in a similar manner, including the ordering of high speed trains: 24 French TGV-derivatives were originally purchased, an order subsequently reduced to 16, but still far more than were needed for the core Madrid-Sevilla service. LAV was ultimately built to International gauge, not Iberian – another reportedly rapid decision – so its trains could not be used elsewhere. The solution was to deploy the spare trains on a bespoke regional service, Madrid-Ciudad Real-Puertollano (a small industrial town on the southern edge of La Mancha) called “AVE Lanzadera” (AVE Shuttle). Thus in just three years, a sequence of “accidents” added a regional layer to a national service, albeit on a distinctly marginal basis: The modern AVE Lanzadera between Madrid and Puertollano, now branded “Avant”, still requires an operating subsidy of 5 million euros per year, roughly €5 per passenger journey, in contrast to AVE services between Madrid and Sevilla, which are operated without direct state support.
As further LAV lines opened in the late 2000s, the Avant (formerly AVE Lanzadera) service concept expanded, with mixed results: Dedicated Avant services (trains carrying only Avant passengers) proved to be a particularly inefficient means of serving small populations, such as Calatayud, yielding inadequate frequencies for passengers and poor seat utilisation for the operator. After rationalisation in 2013, an alternative model became more common, in which seats were reserved for Avant on existing (longer distance) “commercially operated” high speed trains: A theoretically more efficient approach, but also one with theoretical potential for cross-subsidisation between supported and commercial operations. Barcelona-Figueres, which opened in 2013, blurs the distinction between AVE and Avant by offering Avant seats on every AVE service, rendering fare the only practical difference – the train, the frequency, the journey time, the boarding, all functionally identical. The casual traveller might reasonably ask, why not just price AVE and Avant the same? As this essay explores, the fare structure and the funding of each service is quite different, because the formal role of each service is also quite different – but are those differences still valid when the services have become ostensibly identical?
Fares between every Avant origin-destination town pair have been sampled in one direction (only), except in Andalucía, where five pairings provide a broadly representative sample of the Sevilla-Málaga corridor. Only direct trains (single-stage journeys) were included. Track distances are from ADIF and typical journey times from Renfe’s timetable. Historic Iberian gauge lines are normally much slower than LAV, but differ less in distance. There are several notable exceptions: Málaga-Sevilla LAV’s route distance (via Córdoba) is 40% longer than the more direct Iberian gauge line. In contrast, LAV Madrid-Ciudad Real and Madrid-Valladolid are much more direct than the equivalent Iberian gauge lines. In addition to Avant, all town pairs except Madrid-Toledo are served by one or more alternative Renfe product brands, as grouped below. Long-distance products operating solely on slower Iberian gauge lines (Euromed, Talgo, Trenhotel) are not included, since over the few Avant corridors they operate, they are rarely cheaper than regional services, and not as fast as AV:
- AV Commercial – AVE, Alvia, Altaria, AVE-TGV: Operated on the same high speed LAV lines as Avant, with similar journey times, but as part of longer journeys between regions and offering better on-board facilities.
- AV Budget – AV City, Intercity: Like AV Commercial, operated between regions, in these examples on LAV, but with limited schedules (typically one or two journeys per day in each direction) and fewer on-board facilities.
- Regional – Media Distancia, Regional Express, Regional, Cercanías: Operated entirely on traditional Iberian gauge lines, typically much slower than LAV. Generally intended for local journeys, these services may operate between regions, but primarily to serve intermediate stations. For some towns (Antequera, Figueres, Segovia, Tarragona and Utiel) regional services use different (more central) stations to AV services.
Fares, as listed on the Renfe website, were observed in mid-June 2018. Avant and regional fares do not vary by train or day – the only potential limitation on travel is a full train, which is rare, but can occur in busy periods. AV Commercial and AV Budget offer a standard “Flexi” ticket, but both may also offer cheaper “Promo” tickets for specific trains on specific days, as determined by Renfe’s yield management system. To obtain a representative sample of these cheaper fares, all fares in 7 day periods for 0, 1, 2 and 3 months ahead were captured, and the cheapest fare on each day recorded. Two month-ahead data equated to mid-August, a period when work-related travel sharply decreases (for summer holidays). On many of the routes observed, only a small proportion of trains each day were discounted. Passengers that are not able to alter their schedules, will often pay the full Flexi fare: As discussed later, Renfe’s system of yield management is not intended to manage yield over the mid-distances associated with Avant corridors. All data is for a one-way (single) adult fare. Return fares typically offer an overall reduction of 20% (on the cost of two single tickets). Regular travellers using multi-journey tickets can obtain reductions of around half, which is especially relevant to Avant, where 36% of all journeys are made with such tickets. The table, which is conveniently summarised below, shows the raw fare values for each town pair. “AV Cheapest” is the absolute cheapest AV Commercial or AV Budget fare found across the entire 28 days of sampling (7 days at 0, 1, 2 and 3 months in advance), a crude metric which will be expanded further later. The percentage shown is that of the Avant fare.
|From – To||Avant||Regional||AV Commercial||AV Budget||AV Cheapest|
|Córdoba – Málaga||27.50||–||–||41.60||151||27.90||101||13.40||49|
|Málaga – Antequera||10.20||–||–||26.40||259||10.50||103||5.05||50|
|Málaga – Sevilla||43.60||24.10||55||–||–||–||–||–||–|
|Sevilla – Antequera||38.30||18.10||47||–||–||–||–||–||–|
|Sevilla – Córdoba||20.80||13.55||65||30.40||146||21.00||101||10.95||53|
|Zaragoza – Calatayud||14.80||7.85||53||17.50||118||–||–||17.50||118|
|Ciudad Real – Madrid||27.60||28.10||102||37.60||136||28.10||102||18.40||67|
|Madrid – Puertollano||33.50||–||–||41.20||123||34.00||101||20.60||61|
|Puertollano – Ciudad Real||6.90||5.05||73||17.50||254||7.00||101||3.35||49|
|Toledo – Madrid||12.90||–||–||–||–||–||–||–||–|
|Madrid – Segovia||12.90||8.25||64||24.40||189||13.00||101||7.30||57|
|Segovia – Valladolid||19.20||–||–||29.30||153||29.30||153||22.85||119|
|Valladolid – Madrid||29.10||24.10||83||37.30||128||37.30||128||14.90||51|
|Barcelona – Figueres||21.00||12.00||57||36.10||172||–||–||10.85||52|
|Barcelona – Girona||16.20||8.40||52||31.30||193||–||–||9.40||58|
|Girona – Figueres||6.90||4.10||59||22.90||332||–||–||6.85||99|
|Barcelona – Lleida||29.40||13.50||46||37.90||129||29.70||101||8.90||30|
|Barcelona – Tarragona||17.30||7.00||40||28.30||164||17.50||101||5.25||30|
|Lleida – Tarragona||14.00||–||–||20.20||144||17.00||121||5.10||36|
|A Coruña – Santiago||7.35||6.10||83||15.80||215||–||–||15.80||215|
|Ourense – A Coruña||23.95||–||–||25.50||106||21.10||88||10.20||43|
|Ourense – Santiago||16.60||10.00||60||20.10||121||16.60||100||6.05||36|
|Valencia – Utiel||12.90||5.80||45||22.40||174||13.00||101||10.25||79|
To summarise those numbers, Avant is:
- Always more expensive than regional – except Ciudad Real-Madrid, where the Media Distancia fare is 2% higher than Avant.
- Always cheaper than AV Commercial.
- Almost always priced just below non-discounted AV Budget – a relationship explored later in the context of Intercity.
- Often, but not always, more expensive than the absolute cheapest AV (AV Commercial or AV Budget).
It transpires that Avant’s fare structure is extremely consistent (as demonstrated more clearly in the graph below), but there are substantial variations within other fare structures on certain routes. Notably, there is no evidence of any yield management whatsoever between Zaragoza and Calatayud, or A Coruña and Santiago de Compostela, while AV Budget fares to Valladolid offer no discount over AV Commercial. The graph below plots fare by distance, the strongest correlation found, excluding unrepresentative data: The AV Budget plot excludes Valladolid, and the AV Cheapest plot excludes routes on which yield management was found on less than 10% of all days observed – Zaragoza-Calatayud, A Coruña-Santiago, and Valladolid, again. The R-squared values measure how well the data fits the plotted line, where 1 (100%) would be a perfect correlation. AV Commercial and AV Cheapest show the least correlation – the first has no observational error, so reflects the influence of another variable beyond mere distance, while AV Cheapest is especially prone to sampling errors. As detailed later, the regional line conflates 3 separate fare tiers, and thus is drawn on the correct trajectory, but with a relatively low R-squared correlation. For clarity and ease of comparison, all relationships are shown strictly (straight) linear, however Avant fares could also correlate to a power relationship – the lack of short-distance Avant journeys makes this impossible to investigate further. Likewise, overall the lack of short-distance data means the precise intercept values (with the y-axis) are unreliable.
The graph shows the broad relationship of Avant to other fare products. Avant is almost double the regional fare, and, depending somewhat on distance, around €10 cheaper than AV Commercial. However, Avant is well within the range of yield-managed AV fares, which can drop as low as regional fare level. This exposes the potential conflict between Avant fares, which are state-supported as a public service obligation (Obligación de Servicio Público, OSP), and AV fares, which are supposedly not state-supported. OSP presumes no potential provider is willing to offer a service commercially, and while the risk of collision between public and private is minimal at regional fare level, Avant fares are far too close to existing commercial fares to be able to conclude such separation.
Prior fare analysis indicated the maximum extent of discounts on “commercial” (non-Avant) AV fares, but in practice the absolute AV Cheapest metric hides substantial variations, both in the magnitude of discounts and the frequency with which discounts may be offered. Analysis is limited by the observations, which only recorded the best discount on a each day, not the proportion of trains on that day offering that, or another, discount. However, some broad patterns can still be extracted from this limited dataset, as shown in the table below. The first data column lists the percentage of all sampled days on which at least one train was discounted from AV Commercial, where any AV Budget fare (even a standard Flexi fare) that is below the AV Commercial fare is considered a discount. Subsequent columns show the average value of the discount (expressed as a percentage of the AV Commercial fare) of the cheapest train on each day – first for the whole sample, then for each separate period 0 (June), 1 (July), 2 (August), and 3 (September) months ahead. Finally the standard deviation of all discounts is given, which measures the variation within the data – higher values indicate greater variation, less consistency. Average values are presented at the bottom. These simply averages all the data above, without weighting origin-destination pairs by a metric such as patronage, so must be read as crude indicators. “Always Discounted” averages only those pairs where 100% of days in the sample offered a discount. The three sampled journeys with no AV Commercial fare are omitted.
|From – To||% of Days with Discount||Average Daily Discount by Months Ahead||Standard Deviation|
|Córdoba – Málaga||96||45||47||29||48||56||14|
|Málaga – Antequera||100||51||49||49||51||55||16|
|Sevilla – Córdoba||100||40||37||31||33||57||15|
|Zaragoza – Calatayud||0||0||0||0||0||0||0|
|Ciudad Real – Madrid||100||40||25||41||47||46||10|
|Madrid – Puertollano||100||38||21||36||48||47||12|
|Puertollano – Ciudad Real||100||62||60||60||63||67||6|
|Madrid – Segovia||100||61||49||61||67||67||11|
|Segovia – Valladolid||7||2||6||0||0||0||6|
|Valladolid – Madrid||7||4||6||9||0||0||13|
|Barcelona – Figueres||100||56||70||38||44||70||15|
|Barcelona – Girona||50||35||70||0||0||70||36|
|Girona – Figueres||50||35||70||0||0||70||36|
|Barcelona – Lleida||100||56||52||47||55||68||10|
|Barcelona – Tarragona||100||65||65||58||64||74||7|
|Lleida – Tarragona||100||59||49||55||62||69||10|
|A Coruña – Santiago||0||0||0||0||0||0||0|
|Ourense – A Coruña||100||36||25||27||40||50||16|
|Ourense – Santiago||100||50||32||49||56||63||16|
|Valencia – Utiel||61||24||44||4||7||39||21|
|Always Discounted Average||100||51||44||46||52||61||7|
An average discount of 51% was found on routes where at least one train was discounted on every day sampled. The overall pattern matched the common expectation that the further ahead the seat is booked, the cheaper the fare paid – although the discount gained by booking one month ahead is often much the same as a few days ahead. It is the quirks, the journeys with a low proportion of discounts or a large amount of variation, that best help explain Renfe’s approach to yield management:
Fares between Madrid and Valladolid are almost never discounted, however the Alvia trains on this route are yield managed, just for longer distance journeys. This commonly inverts the expected relationship between price and distance: Longer distance (discounted) fares from Madrid to Burgos, or even further to Bilboa, can be cheaper than the (non-discounted) fare from Madrid to Valladolid, even on exactly the same train. The pattern is indicative of a “yield management” algorithm that is only managing a part of the yield: For each train, apparently selecting a subset of origin-destination pairs to be managed, while the remainder are left undiscounted, perhaps regardless of the train’s eventual occupancy. The subset selected can be remarkably consistent from week to week, albeit with set seasonal or weekend (Friday/Sunday) variations. For example, some of the AVE trains on the Barcelona-Figueres corridor (including Girona) serve a notoriously peaky summer tourist demand to/from France. This helps explain why in June and September (+0 and +3 months), certain trains offer consistently high discounts across the week, but in July and August (+1 and +2 months), these discounts drop dramatically, to no discount at all to/from Girona.
Overall, Renfe’s current yield management is clearly focused on longer-distance journeys, and on many Avant corridors the yield management observed is almost incidental: Primarily an attempt to fill under-used long distance trains, not specifically an attempt to manage journeys within regions. The only corridor where shorter distance journeys appear to be actively managed, with a high proportion of all AV Commercial trains discounted, is Barcelona-Lleida. The most obvious explanation is that Barcelona-Lleida is part of the busiest and most contested (by aviation) long distance corridor in Spain – Madrid-Barcelona – and thus the corridor Renfe are most motivated to optimise. At least between cities – Zaragoza-Calatayud is on the same corridor and apparently never yield managed. This makes assessment of the potential for public-private collision on price, between supported Avant and commercial AV, difficult to assess: Does Renfe’s apparent disinterest in regional yield management imply a more commercially-orientated or technologically-capable competitor would be similarly disinterested? Is there a fundamental cultural reluctance, be it by operator, customer, or state, to accept yield management’s inconsistency within regions? Some insight into these questions can be gained by examining approaches to non-AV fares within regions.
Paying to Go Faster
Aside from high speed Avant, there are four other Renfe fare products associated with the public service (OSP) network: Cercanías, Regional, Regional Express, and Media Distancia. Cercanías is specific to large urban areas and their immediate travel-to-work hinterlands, and commonly uses a simple zonal fare structure. Core Cercanías services have been excluded from this analysis because their focus is generally on an urban area, not an Autonomous Community region. Where the urban area is the region, specifically Madrid, the fares within have not been analysed. 264 origin-destination journeys were sampled across the Renfe network, excluding Feve (primarily the metrica-gauge route along the north coast) and lines owned by Autonomous Community governments (primarily in Catalunya and the Basque Country). Only routes with at least one Regional, Regional Express, or Media Distancia fare were included. Sampling focused on main regional towns, with a mix of journeys between these towns, and journeys to intermediate stations – the aim to reflect the full spread of fares, but with emphasis on those fares most likely to be used. Adult one-way fares were again taken from Renfe’s website, distances from ADIF’s documentation. Avant data is as described previously.
383 fares were sampled, in addition to Avant data. As with Avant fares, the best correlation was (straight line) linear with distance. The true pattern is not strictly linear, but stepped, with fares according to distance bands. The fare structure occasionally conflates fares, most notably in Galicia, where Regional and Regional Express fares can be identical, as can Regional Express and Media Distancia – as if a national structure is trying to accommodate just two fare bands. A similar conflation occurs with Avant and Media Distancia in Galicia (A Coruña-Santiago de Compostela), where it presumably reflects the unusual style of MD operation on the A Coruña-Vigo corridor: Media Distancia operates with similar trains and speed to Avant A Coruña-Ourense (on average 109 km/h to Vigo vs 124 to Ourense), and until recently was operated as “Avant”.
Only one group of exceptionally long distance journeys, categorised here as Madrid-to-Periphery, did not adequately fit the common patterns shown in the table below over the longest distances. This group includes the daily Madrid-Barcelona “Regional Express” that ostensibly serves the 40 largest intermediate stations along 659 kilometres of now largely historic line between Madrid and Barcelona – the train’s total journey duration more than three times that of the parallel AVE. Similar services operate from Madrid to Valencia, the Basque Country and the far south-west (Badajoz, Huelva). Except for Extremadura, these services are routinely surpassed by faster alternatives for long distance railway journeys to Madrid. Madrid-to-Periphery’s lower fares (on average 15% below the patterns elsewhere) at far longer distances (an average sampled journey length of 512 km, compared to 93 elsewhere) suggest the fare-distance relationship may not be strictly linear. However the limited opportunities to sample such long distance fares, and the little relevance such journeys have to genuine “regional” travel, makes it pragmatic to exclude these journeys from the analysis.
|Product||Sampled Journeys||Average Distance of Sampled Journeys (KM)||Derived Fare Equation||R-squared|
|Regional||88||81||0.073 * distance + 1.042||0.96|
|Regional Express||128||91||0.082 * distance + 1.005||0.98|
|Media Distancia||150||98||0.098 * distance + 1.063||0.97|
|Avant||22||122||0.149 * distance + 1.945||0.99|
In broad terms, Regional Express is over 10% more expensive than Regional, while Media Distancia is almost 20% more than Regional Express or roughly a third more expensive than Regional. Avant is double Regional. All these services are supported by a (national) state that routinely alludes to equality – contemporarily in the context of territorial cohension. The prevailing explanation for these price differences is that the cheaper services have more stops, thus travel slower, and sometimes use older rolling stock, thus tend to lower quality. Herein a reference to the title of this essay – “Is Alta Velocidad Fast?” – that traditionally, faster trains have carried a higher price, but that structure may be threatened by modern AV.
Except for services wholly within Catalunya, whose 2006 Statute exerted greater Autonomic control over railways than elsewhere in Spain, the regional trains provided by Renfe as a public service are contracted by the Spanish government. The only current public data for these services is that analysed by INECO in 2017, primarily using 2015 performance data (a reprise of a similar 2013 report that used 2010 data). Known quirks, especially in eastern Galicia (discussed later), have been adjusted to improve the quality of the data. The Córdoba-Universidad Rabanales service has been excluded because while technically it is Regional, its relatively high frequency, 7 urban kilometre route has the character of Cercanías. Performance metrics for each route were distributed in proportion to the number of train journeys of each product (Regional, Regional Express, or Media Distancia) offered each week on the route. Only 12% of the routes analysed offered more than one product, so in most cases no factoring of data was required. The table below summarises the distribution between products. Train-kilometres are the distance operated, passenger-kilometres the distance travelled by customers. Regional trains tend to be associated with much shorter journeys (both for trains and, especially, passengers). The three slower (non-AV) tiers are strongly skewed towards Media Distancia, even in the count of trains (one-way end-to-end train journeys).
|Product||Proportion of All||Averages|
|Trains||Passengers||Train-KM||Passenger-KM||Passengers / Train||Train KM||Passenger KM|
The average scheduled operating speeds (end-to-end) for each product on each route, were weighted by the percentage of the all (separately, train-km and passenger-km) of that product operated on each route. The results were summed to create an average speed by product, weighted by either train or passenger distance. As seen in the table below, Regional Express are indeed faster than Regional, while Media Distancia are faster than Regional Express – both train and passenger weighting showing similar patterns (% R is the percentage of the Regional value). Although overall speed accords to product, there is considerable variation between routes, which is not conveyed by the averages in the table. For example, Zaragoza–Huesca Regional services average about 90 km/h, well above average, while Zaragoza–Valencia Media Distancia services are closer to 70 km/h, well below average.
|Product||Speed (train-km weighted)||Speed (passenger-km weighted)||Fare at 117 KM||Fare at Average Passenger KM|
|km/h||% R||km/h||% R||€||% R||€||% R|
These speed differences are remarkably well matched to the fare structure by kilometre: Each fare formulae (derived earlier) has been applied to the average passenger distance (over all analysed services) of 117 kilometres. Only the Avant fare is out of proportion to its speed – 202% of the Regional fare, but offering 233% or 228% of the speed – the implication, that in order to maintain the relationship to speed found among the slower products, Avant fares should be about 15% higher than their current prices. Since the average passenger distance varies by product, the fare at the average passenger-distance for each respective product (final columns of the table above) has a greater spread, bands of revenue per passenger that increment 50% of the previous, from Regional through to Avant. So while an operations-lead railway culture could reasonably price services by their operating speed, an accountancy-lead culture could just as reasonably group products by their revenue generation.
Annual revenues and costs from 2015 have been extracted from INECO’s report, initially with just one adjustment – Madrid-Sigüenza, which equated to €63/train-km, ten times the cost of similar routes, and has been reduced accordingly. While, as shown in the table below, revenue per passenger is indeed higher on faster products, the revenue per passenger kilometre is reasonably consistent across the three non-Avant products. And again, with the exception of Avant, the cost per train journey is higher for faster products (primarily because such train journeys travel further), but the cost per train kilometre is relatively consistent.
|Product||Revenue (euros)||Cost (euros)||Revenue per 000 passengers / Cost per train||Revenue per 000 passenger-km / Cost per train-km|
|Annual (millions)||per passenger||per passenger-km||Annual (millions)||per train||per train-km|
The comparison of revenue to costs produces another set of relatively consistent metrics: Almost as if the purpose of the entire fare structure was to normalise the balance between costs and revenues across the network, in a manner that allowed the flexibility to offer different styles of operation on different routes – and without having to be locally concerned that, overall, Regional covers a lower proportion (27%) of its costs from revenue than Media Distancia (46%). Avant, as a relatively new service concept, has perhaps not matured enough to reflect such a pattern. Of course such patterns could be coincidental, and Renfe’s prime motivation has actually been to operate whatever services it can within a target cost per kilometre. In such a context Avant merely raises the question of how that cost is measured, a question which itself asks whether Renfe’s contemporary focus is on operations or customers: Avant may have the highest cost per train-kilometre (as shown in the table), but it offers the lowest cost per passenger-kilometre (€0.15/km, compared to Regional’s €0.26/km). Overall it does indeed cost the passenger more to travel faster – although it does not necessarily cost Renfe more to operate trains faster, and perhaps there is more to this equation than first appears.
The proportion of each Regional, Regional Express, or Media Distancia route in each Autonomous Community was estimated, and train performance data distributed accordingly. To obtain a robust basis for comparison, Communities were grouped where individually they represented less than 5% of total (national) train-kilometres: Murcia & Valenciana, and Navarra, Pais Vasco & La Rioja – both groups neighbours that share some services. Several communities could not be analysed:
- Asturias and Cantabria – each of which has only a couple of applicable trains each day (to/from Castilla y Leon).
- Catalunya – where the majority of applicable services are not contracted by the Spanish government and thus not in INECO’s data.
- Madrid – where almost all the applicable train-distance is operated to serve Castilla y Leon or Castilla La Mancha, and does not portray Madrid herself.
As shown in the table below, average operating speeds (weighted by train-kilometres) vary by up to 20% between different Autonomous Communities: Aragon and Murcia/Valenciana are notably below average, La Mancha and Extremadura notably above. Although Catalunya cannot be analysed using this method, a more limited analysis using the fares sampled earlier, suggests a balance of products very similar to Aragon (with a strong skew towards the cheapest products). These patterns broadly accord to physical geography – for example, the flat open plains of La Mancha, with few towns and infrequent stops, naturally afford a high average speed. Differences in physical geography are important to the historic (non-LAV) network because 19th century engineering could not simply tunnel underneath any mountains in its path (as is more common in LAV construction), so rough terrain tends to be associated with railway lines that are characterised by their curves and gradients, both of which reduce maximum speed. Likewise the frequency with which local trains stop at stations, and thus become slower over their entire journey, tends to be greater where the population density is higher. It was thus inevitable that pre-AV Renfe would have to manage a network on which operating speed could not reasonably be equalised – be seen to be fair – across Spain. Based on comparison of average speeds and the balance of products in each Community, Renfe may well have managed these differences through its fare structure: For example, Aragon and Murcia/Valenciana have slowest trains, but the fares on those trains are strongly skewed to cheaper products. Likewise speeds are higher in Castilla and Extremadura, which tend toward the more expensive fares.
|Community||Average Speed||Train-km by Product||Cost (euros)|
|Km/h||Variation from All||Total Annual (millions)||Regional||Regional Express||Media Distancia||Total Annual (millions)||per train-km|
|Castilla y Leon||85||105%||5.7||8%||23%||69%||48.0||8.37|
|Castilla La Mancha||91||111%||3.4||19%||18%||63%||35.1||10.19|
|Murcia & Valenciana||71||87%||2.0||37%||27%||37%||20.1||10.20|
|Navarra, Pais Vasco & La Rioja||78||96%||1.4||0%||71%||29%||10.1||7.06|
Andalucía is the only community that does not match this pattern – almost every train is operated as Media Distancia, yet speeds are merely average for Spain. The closure of the line to Granada (and the associated replacement of trains by buses) may have lowered the average slightly – of the four busiest routes within Andalucía, the only route that passes through Granada, Sevilla-Almería, is currently scheduled with a speed well below average. However this one line could not have reduced average speeds across the whole region by 10 km/h – the difference between Andalucía and Castilla La Mancha, the community with the fastest average speed. Andalucía was historically notable as the only region where the counter-balancing duopoly of the Norte and MZA had little influence, the regional Compañía de Ferrocarriles Andaluces dominating until the nationalisation – although that would suggest a century-old fare distortion somehow remains, so perhaps there is another explanation.
Cost might reasonably be expected to be the main determinant of fare, but as seen by comparison with the cost per train-kilometre (in the table above), fares are far better correlated to speed than cost: For example, the lowest operating costs are €7/train-km in Navarra/Pais Vasco/La Rioja, against €12 in Galicia – yet fares in both areas are skewed just slightly toward the cheaper products, while average speeds in both are similar, and similarly just below average. The low costs in Navarra/Pais Vasco/La Rioja may in part reflect the dominance of electrification there – not because electric operation is necessarily much cheaper, but because its “fuel” cost occurs via track access charges, the costs of which have historically been skewed – in comparison, diesel fuel is logically associated to the cost of train operations. The conjecture serves to demonstrate one of the many difficulties in assessing railway costs. Intriguingly, much of the excessive cost in Galicia can be attributed to the Vigo-A Coruña corridor, including the fast “Media Distancia” (neo-Avant) service, whose €16.60/train-km is the highest such cost of any core (those with more than 100,000 train-kilometres per annum) non-AV route analysed. However, the relevant published data is missing about a third of the service’s expected train kilometres, describing an average of 96 trains per week, against the 136 scheduled (as listed in both INECO’s service description, and the current timetable). The parallel slow service is not under-reported. Is a third of the Vigo-A Coruña fast service simply not OSP? And if so, could the cost data still refer to the entire service, including the missing train-kilometres? €16.60/train-km is more typical of comparable dedicated Avant services, such as Madrid-Toledo, although compares favourably to the €22/train-km attributed to Ourense-A Coruña Avant. However, while a part of the Vigo-A Coruña line is maintained by ADIF-AV, whose infrastructure access costs tend to be higher than its Iberian-gauge partner ADIF, the line is too slow for track access to be charged at an AV premium.
That time has a monetary value, which is reflected in consumer choices between competing modes, is well established in transport economics. However such competitive rationale appears quite irrelevant to Renfe itself:
- There are very few routes on which these price differences represent a rational economic choice to Renfe’s own passengers. Even where Renfe offer more than one product (12% of routes analysed), low frequencies mean exercising choice typically involves travelling in the morning rather than the afternoon – nonsense for relatively local journeys: Most passengers will take whatever category of railway service they are offered, or switch modes.
- There is scant evidence that Renfe are adjusting the allocation of products to reflect the competitive environment on specific routes. For example, in spite of years of poor railway patronage between Ferrol and A Coruña, not least because the bus is faster, Renfe still maintains the highest (Media Distancia) fare tier on the route. Nationally, the sometimes substantial differences in fares between regions can be hard to rationalise against other modes: While there can be small differences, for example in fuel costs, it is not obvious that driving a car in La Mancha costs 20% more than in Aragon. More broadly, there is no reason to suppose OSP fares should be competitive with local alternatives, since the public service network is justified largely without specific reference to other modes (not least in INECO’s 2017 report).
Fare differences can be better understood as compensation: The state (traditionally as RENFE) transferring money to counter the deficiencies in the state’s provision – deficiencies compared to Media Distancia, or perhaps now compared to AVE. The monetary transfer an attempt to balance inequalities, to acknowledge that one Spaniard is being offered a “better” service than another Spaniard. As explored with the previous table, balancing areas with different average train speeds has logically always been Renfe’s problem, and it therefore already has a method of balancing the contemporary inequalities between those places with AV and those without. Unfortunately there are modern complications:
- Spanish government policy presumes equality in a future network, largely based on the Aznar-era vision of LAV to every regional capital of Spain. In practice the slow (or possibly never) delivery of LAV across Spain implies compensation while the network is being built. The mere ilusión of a future “AVE” is clearly insufficient in regions like Extremadura, who have already spent 25 years watching their neighbours in parts of La Mancha travel almost twice as fast to Madrid.
- Aznar’s vision guaranteed nothing for many of Spain’s smaller towns, and thus implicitly framed AV as a service between, not within regions. Most local journeys were always going to require some form of equality-balancing compensation. Originally it might have been argued that AV was defined by a remote (international) aviation market. However the requirement for compensation became undeniable once LAV started being used for journeys within regions (AVE Lanzadera and later Avant), and once LAV started to be populated by stations serving small towns (not just the large cities it was conceived to link), because AV could not longer be differentiated as strictly inter-regional.
- The application of commercial rationale (especially yield-management) to inter-regional AV has limited the scope to balance inequalities. For example, if offering regional fares that are only half AVE is not enough to affect a balance, the only price option is to further subvent (to reduce by direct subsidy) regional fares, because AV fares cannot be increased without undermining AV’s competitiveness against aviation.
That final issue primarily arises because “commercial” AV has actually been subsidised through capital investment (especially in LAV construction), such that its operations can match fares set by aviation markets. The underlying structural problem is the low cost of aviation relative to rail, and more specifically the apparent inability of the national government to rebalance, by means of increasing aviation prices, the international aviation market. While all that AV subsidy may be justified as a state intervention to counter the externalities of aviation (pollution, noise), it would have been fiscally more sensible to increase the price of aviation by taxation, and let AV cover its own costs: The primary reason unrelated to the higher classes of transport actually being taxed or subsidised, rather that the effect of subsidising one class of transport cascades onto all lesser classes, and in this case almost all transport classes (in Spain) may be considered lesser classes. Having elevated the benchmark for “better” by (capital) subsidy of AV, “better” is now out of proportion to the traditional railway fare structure. With AV (speed and fare) effectively fixed by the aviation market, the only options are to improve the lesser classes of transport, or lower their fare. The cost of LAV is thus not just its own cost, but also the cost of rebalancing the lesser classes of transport. The strategic moral is surely to subsidise the most local and tax the most national – but sovereign states, perhaps by the very nature of their construction, tend to emphasise the opposite.
Avant, especially where it matches AVE seat-on-train, should provide a basis for establishing the equitable difference in price between AV and non-AV for travel within regions. The previous section’s analysis suggests Avant’s current premium, roughly double Regional, is insufficient to reflect its greater speed. Assuming AV fares are fixed, all three non-AV fare tiers would need to drop by about 20% – or (long term) operate proportionately faster. Unfortunately the Avant network is limited and disproportionately skewed to the biggest cities, its service increasingly somewhere between AVE and fast Media Distancia, and its market far from mature: There are many reasons why Avant may not be representative of an “AV premium”. For example, Avant does not necessarily have a market, even at current prices: Requena and Utiel’s local Cercanías fare to Valencia is 45% that of Avant, but in spite of Avant travelling the route in a quarter of the time of Cercanías, Avant conveys less than 20 passengers each day. While Requena-Utiel has its own quirks (a remote station, with a comparatively limited frequency), the example indicates that Avant should not be assumed interchangeable for slower regional services. AV may have its own market, and that market may be limited to a few busy inter-urban corridors. Away from those corridors, Avant and Media Distancia fares and services are likely to converge – and using the template of Vigo-A Coruña, converge at the Media Distancia fare, rendering no tangible “AV premium” whatsoever.
The Madrid-Barcelona Regional Express offers a further counterpoint: A daily train priced at €42, between a half and a third of a typical AVE fare, but with over three times the journey duration and none of AVE’s quality. On average each train carries 470 passengers, totalling almost a thousand per day (both directions), but each passenger travels an average of just 130 kilometres – a fifth of the total train journey distance – suggesting very few, if any, use the service for long distance travel from Madrid to Barcelona. For context, AVE carries in the order of 10,000 pasengers per day between Madrid and Barcelona (both directions combined, city to city), ten times more passengers than board the Regional Express at any station on its route. It might be useful to conduct price trials on these slow Madrid-to-Periphery trains to ascertain the fare at which passengers started to switch from AVE. Not because the 10 hour Madrid-Barcelona market is ever likely to be viable for a railway operator, but because it would help define the true – equitable – cost of serving Spain with such a diverse spread of railway services.
For the analysed non-AV regional services, the cost of supporting fare subvention to offset inequalities might not be excessive, certainly not excessive in the context of the LAV construction it would counter-balance: These trains only cover 41% of their current costs from revenue – even halving their fares would only cost at extra €60 million each year, equivalent to extending LAV by just 5 kilometres. While such reductions would logically generate additional passengers, overall these trains convey 3.6 times more seats than are used. While that metric is misleading – it includes seats offered to satisfy demand elsewhere on the route, which cannot reasonably be filled throughout the train’s journey – the sheer magnitude of excess seats suggests much of the existing network has capacity to grow its patronage before requiring many additional assets or significantly increasing operating costs. The flaw in such fare subvention is not financial, but its potential cascade effects, for example onto parallel bus operations, cascading to local taxis, car sales, bicycle maintenance, and so on.
It is perhaps only once the same logic is applied to local urban services, that properly counter-acting AV‘s inequalities through fare subvention becomes more obviously infeasible: A large proportion of all Spain’s railway passengers are carried by Cercanías, Renfe’s local suburban services, almost always at the lowest tier of speed and quality. The ultimate, politically unanswerable question: Why isn’t every Cercanías passenger entitled to the same speed and quality the state affords to AVE passengers? While there are technical answers to that question – infrastructure construction can be far more expensive in urban areas, acceleration between frequent stops inhibits the scope for high speed – there are also societal reasons: Functionally, that the (commercial) elite need to travel more than the (working) plebiscite that those elite support. Politically, that the territorial state needs to connect itself in order to function as a stable political society. Both may be increasingly historic in their importance, but are still factors in modern Spain. Both factors also create, and are thus ultimately counter-balanced by, societal inequality – the combination acts as a dynamic counter-balance within “state”. In Spain, the fluid functioning of that balance is complicated by both the:
- Tendency to extremes, which means a balance can grow extremely unbalanced before its eventual readjustment: The contrast of AV and non-AV railways provides a good example – one perfectly modern and generally over-specified, the other too often a relic, only sporadically rationalised or modernised – with aparently little enthusiasm for the intermediate position of merely modest improvement. This pattern is deeply rooted in the societal model of hope.
- Tendency to mask underlying inequalities within structures that superficially appear equal, which makes it difficult to comprehend imbalances on a national scale: Here the same national structure appears to apply to all fares within all regions, so must be fair – except, as this essay explores, the structure both masks and balances unseen inequalities. This pattern reflects the voids in transactional responsibility between groups built on the societal family model.
Functionally, Avant is the closest actual service to that theoretical model of a high speed, high quality, “AV Cercanías”. State support of Avant is, in part, officially justified by the amount of passenger “recurrence” – commuter or similarly regular travel. Much has been written about the “economic and social” benefits of Avant, particularly to the smaller places it serves, such as Ciudad Real. In transport geography terms, such places are becoming the outer suburbs of conurbations like Madrid, while retaining the cultural nuances and sense of place that such outer suburbs can easily loose: Toledo shares the same (35 minute) railway time relation with Madrid as Parla, the suburb that Toledo’s Avant trains speed past without stopping – even though, in physical distance, Toledo is three times further away from the centre of Madrid than Parla. For regular passengers, Avant reasonably becomes a part of their internalised “family” (home, work, life) geography, and it is in this regard that Avant differs fundamentally from AVE: As discussed in The Expectation of Competition, LAV provides “presence” – the physical manifestation of authority, theologically rooted in the state’s manifestation of God in nature – AVE akin to a rarely visited cathedral. The equivalent Avant metaphor would be a priest hearing Confessions – a state that is far more deeply integrated into the being of its users – and thus not a service that can be curtailed without strife.
However for many commentators Avant evokes sentiments of inequality, typically characterised as “despilfarro”, that the government is not spending shared resources fairly. Avant is thus pushing a model with severe underlying inequalities, to the point where those inequalities might even become visible, which would presumably herald a societal rebalancing – albeit a challenging change, because for some members of this society, Avant has become firmly embedded in their lives. However, if Galicia is any guide, the solution to Avant’s inequalities will simply be to merge it into the Media Distancia fare structure, thereby creating far greater inequalities overall: A neo-AVE that costs just a third more than a Regional train – a Regional train whose journey time may be two or three times longer, and whose rolling stock may be two or three decades older. Equality be damned?
In 2012, in the midst of post-Crisis financial austerity, the Spanish government sought to reduce the cost of Renfe’s operations, primarily by rationalising inefficient services (inefficiency measured as a combination of patronage and operations). Some of the more notorious railway excesses had already ceased, such as the Toledo-Cuenca-Albacete “AVE of La Mancha”, reportedly requiring an operating subsidy of over a thousand euros per passenger – although the expectation of such a service remains. None of the proposed line closures happened (except Córdoba-Bobadilla, which roughly parallels Avant between Córdoba and Antequera), and while Renfe still muses withdrawing passenger services from the old (non-AV) line to Segovia, the momentum has largely reversed as Spain’s economic conditions have improved: The Burgos “direct” line, proposed for closure and in any event closed by the collapse of a tunnel in 2012, is now actively proposed for reopening. Perhaps the most tangible outcome of this period of “rationalisation” was to evolve the means of local service delivery – specifically to merge the operation, and to some degree financing, of regional and longer-distance services – a structure explored in this section.
On some routes seats specifically for local journeys are offered on long-distance (inter-regional) trains, typically trains operated “commercially” under AVE or Alvia brands. These seats are sold as regional OSP fare products (Regional, Regional Express, Media Distancia, or Avant), but are only sold for local origin-destination pairs. For example, the morning train from Gijón to Barcelona is labelled Alvia and operated using high quality AV-capable Series 130 rolling stock, whose capital cost per seat is more than double that of rolling stock specifically designed for local services. Fares are otherwise yield managed – a complete (adult, one-way) journey anywhere between 27 and 69 euros. For journeys between Gijón and León the same train offers Regional fares (for example, €13.35 for Gijón to León), and between Vitoria and Zaragoza, Regional Express fares (€22.15 for that particular pairing). On both sections the train operates on the slower Iberian-gauge network, and thus while the quality of the train’s composition is exotic, its speed is unremarkable. However, for exactly half its duration (5 hours 30 minutes, from 11 hours overall) this train is forfilling a state-funded public service obligation, not just the commercial long-distance function the whole route offers. What may be presented as a marginal use of seats on an otherwise commercial train, can just as easily be interpreted a lot less marginally, theoretically as two OSP route segments joined with two non-OSP route segments – the associated costs of each provision potentially difficult to audit.
In 2012 Renfe revived their Intercity brand, officially for long-distance services operated like Media Distancia services – typically using a lower quality of rolling stock than AVE or Alvia, no premium (Preferente) class seating, no proper catering, and tending to operate more slowly overall – although segments of Intercity services do use LAV at regular AV speed. A historical review of the first 6 years of this new Intercity product suggests there have already been many changes to the service pattern, and the current role of Intercity is certainly inconsistent: For example, one Intercity service from Valencia to Madrid takes the slower pre-LAV line via Alcázar de San Juan, while the remainder of the Intercity service operates over LAV, merely extended beyond Valencia to towns that are otherwise only served by Valencia’s Cercanías. The Intercity service between Barcelona and Pamplona operates no more than once per day in each direction, largely mirroring the prevailing Alvia service except between Lleida and Zaragoza, where in one direction only the Intercity journey can take 30 minutes longer without serving any additional stations (presumably switching off LAV and onto the historic Iberian gauge line, for no obvious passenger benefit). Over the rest of its route the train carries OSP seats – “Avant” seats between Barcelona and Lleida and “Media Distancia” seats between Zaragoza and Pamplona. While these overlaps between commercial and OSP operations are common on Intercity services, they also occur on other products, such as the Gijón-Barcelona Alvia described above.
The defining characteristic of Intercity is price: While products like Alvia may continue to offer full “commercially priced” fares for longer-distance journeys on trains that are locally shared with public service obligations, Intercity typically offers a consistently low fare throughout. And since Intercity fares are yield-managed, their fares tend to remain comparatively low in all situations. Many basic (non-discounted) Intercity fares are set just above the equivalent regional fare (a regional fare often offered on exactly the same train), yet yield management often drops the Intercity fare below that regional fare, making the Intercity fare the absolute cheapest. How Renfe are able to offer such low fares on Intercity services? The modern vudú of yield management provides an inadequate explanation: For ease of demonstration, assume the train-kilometre-weighted averages for the (non-Avant) services analysed in previous sections – a 117 kilometre journey, 208 seats per train, €9.89/train-kilometre. At full Regional fare, at full capacity throughout, one train journey would generate €1993 and cost €1157 to operate. The vast majority of passenger can be assumed to be making return journeys, for which a net 20% reduction is offered by Renfe, reducing revenue to €1594. The maximum theoretical profit margin is thus just over a third of operating cost, which is the margin with which to offer discounts, or with no discounts, defines the minimum seat occupancy as no less than 70% over the entire route. Reconciling those margins would be challenging between Madrid and Barcelona, but surely quite improbable on routes of a relatively local character, with many different passenger flows, each unbalanced by dint of the underlying population distribution and time of day: For example, try matching the loads in the first hour of Barcelona-Pamplona (18:40 Barcelona-Lleida – dense population/commuter peak) with the final hour (21:53 Castejón-Pamplona – sparse population/late evening). AV equivalents are set at a higher fare level (just above Avant), but their operating costs are also higher: The five dedicated (not shared) Avant services average almost €19/train-kilometre, so even at double the fare of Regional, the AV revenue-cost balance is rather similar to non-AV.
In 2014 the Intercity concept was extended to services operating solely on LAV as AVE City, later “AV City”, promoted with the marketing inglés of “low cost”. The phrase better translates as long-distance Avant, albeit only at the margins: The first and last trains of the day between Sevilla and Madrid. Valencia-Málaga at the weekends. Daily Madrid to León, and since May 2018 to Zamora, towns currently served by incomplete fragments of the intended LAV network. The prime characteristic of contemporary AV City is its provision of the services that commercial AVE shuns. Any pretension to develop a coherent network of low cost alternatives to AVE has been forgotten – perhaps in favour of “EVA”, the alternative AV service Renfe announced at the start of 2018. As with Intercity, AV City is priced at just above the equivalent regional fare, here Avant, and yield-managed such that the fare may drop below. However the overlaps with Avant are more subtle – operating over Avant routes, apparently with Avant rolling stock used on a marginal basis – but only rarely (Valladolid-León) offering OSP seats.
Many of the train services that share OSP and commercial roles are difficult to identify. For example, the early morning Madrid-Gijón Alvia is shared with Media Distancia between Valladolid and Gijón – but only on Saturday – on weekdays exactly the same journey is priced “commercially”, typically 50% higher. The associated performance data may, or may not, have been aggregated into the dedicated (not shared) daily Regional Valladolid-Gijón service: Unfortunately INECO’s report significantly under-states (at 241 kilometres) the route length, rendering subsequent possibly-factored-possibly-not performance data unreliable. Elsewhere INECO demonstrate the difficulty of auditing such services more admirably: For example, (in tabla 72) they report 0.3 million annual train-kilometres with an average (derived from seat-km/train-km) capacity of 36 seats (fewer than Renfe’s single-car Tamagochis, all-but confirming at least some seats on shared trains) under the nebulous heading of “Vigo-Valladolid”, a grouping for which they (unusually) provide no train-by-train service description. The core twice-daily Vigo-Ponferrada-León Regional Express uses a dedicated train with many more seats, accounted as Vigo-Ponferrada and Ponferrada-León. The Vigo-Valladolid route is only served in its entirety by the overnight Trenhotel, which does not directly offer regional fares, although from Ponferrada to Vigo (and from Ponferrada to A Coruña) Trenhotel operates with Intercity seats on a route where other Intercity trains do provide regional seats. Most logically, “Vigo-Valladolid” summarises several shared services within the corridor – perhaps Vigo-Ourense and León-Ponferrada on Intercity, Ourense-León on Alvia, and who knows what on Trenhotel. The best match to “Ferrol-Ourense” appears to be Lugo–Ourense, where one daily Media Distancia service is provided on an Alvia bound for Madrid, however this service fragment only accounts for half the train-kilometres INECO have allocated to “Ferrol-Ourense” – their calculations suggesting the whole route is served. Querying these missing service descriptions might be considered pedantry, were the two groups combine not requiring an annual subsidy of 1.1 million euros, support justified to serve local mobility needs that cannot be located locally.
The implication of INECO’s difficulties is that Renfe’s accounts are operational, making no distinction between public and commercial at local route level. Such an operational accounting structure does not just raise problems of transparency where “commercial” trains carry OSP seats for a part of their journey. It means many of the reported metrics will necessarily be factored – a proportion of a shared value, which could be apportioned in all manner of ways: For example, is the operating cost of 20 seats on a 200 seat commercial train, 10% of the commercial train’s cost? Or perhaps what it would have cost to operate a separate regional train, maybe 80% of the commercial train’s cost? Or maybe the potential revenue lost by not being able to sell the seats for journeys to Madrid? Assuming that overall the performance data is correct, even where its precise allocation is unreliable, services offering an average of under 50 seats per train can easily be identified as shared, and these form the basis of the analysis in the table below.
|Service/Corridor||Train-km (millions)||Seat-km (millions)||Cost (million euros)||Cost / train-km (euros)||Cost / seat-km (euros)||Seat Occupancy|
|Miranda de Ebro-Valladolid||0.147||0.005||0.499||3.39||0.094||34%|
|Total/Average listed services||1.120||0.040||3.774||3.37||0.096||43%|
|Total/Average all analysed Non-Avant||27.46||5.71||271.5||9.89||0.048||28%|
|Total/Average listed services||0.428||0.011||2.605||6.08||0.242||49%|
|Total/Average all Avant||7.24||1.62||115.9||16.01||0.072||49%|
On non-Avant services the cost per train-kilometre is very consistent, at a third of the average cost across all the analysed services. The cost per seat is double the average, but roughly equals the (Media Distancia) passenger fare for the seat. Following the principle that the government matches the financial contribution of the passenger (an equation which characterises the overall balance of revenue and subsidy), the price paid by the government “is fair”, and the arrangement certainly appears cost-effective. Except seat occupancy averages less than half: The government is effectively buying a lot of typically unused capacity – in the worst case above, Almería-Linares Baeza, 80% of the 24 seat-capacity allocated for regional travel on the daily Almería-Madrid “Talgo” service is under-utilised. While anywhere close to 100% occupancy is unrealistic on most local services, and overall the seat occupancy on these service is well above average, there is perhaps scope to better optimise the number of reserved seats. But while the government’s (effective) purchase of these regional seats will tend to enhance the overall commercial viability of the associated trains, there is no evidence here of wilful cross-subsidisation.
That is less easily concluded of Avant. Zaragoza-Calatayud follows the non-Avant pattern, albeit all the monetary values are almost double (as expected, given Avant’s higher fares and operating costs), but Valencia-Requena/Utiel defies such rationale: Aside from a seat offer that is intrinsically hard to match to such low demand (18% occupancy of 11 seats per train is just one passenger), the cost of each seat (regardless of its occupancy) is excessive – higher than the full AVE Preferente fare, only rational if each seat’s short-distance reservation prevents much greater revenue being earned on longer journeys toward Madrid. Yet Valencia-Requena/Utiel’s cost per train-kilometre remains remarkably consistent with that of Zaragoza-Calatayud. Cost per train-kilometre is similarly consistent for the non-Avant services listed, raising the conjecture that the train is being subsidised, not specifically the seats. Further investigation of other known shared services reaffirms this tendency to normalise the cost per train-kilometre, with scant regard for seat occupancy. Could OSP seat reservations on commercial trains primarily be designed to suit operational – not specifically passenger – demand?
Analysis of all services on the Pamplona-Zaragoza corridor suggests INECO’s (Tabla 78) data for “Pamplona-Zaragoza” actually just describes the service specifically between those points: Currently, a daily Regional Express train (in each direction), and 6 days per week, one Media Distancia – that second service shared with the Pamplona-Barcelona Intercity as far as Zaragoza. Other OSP services on the corridor fall under other route headings. The combined pairing averages 242 seats per train, operating at an unremarkable €8.50/train-kilometre, covering half of that cost from revenue. The Regional Express is generally reported to be operated by a 3-car Series 470 unit with a capacity of 220 seats. To maintain the route average, the Media Distancia service would therefore need to offer 270 seats. The Pamplona-Barcelona Intercity is reported to use Series 120-derived rolling stock, initially a mix of 120 and 121, although current references point to a single Series 121 unit, with 282 seats. Assuming these rolling stock allocations are correct, the implication is that almost all the Pamplona-Barcelona Intercity’s capacity between Pamplona and Zaragoza – the most difficult capacity to fill commercially based on population density and time of day – is classed as Media Distancia and supported as OSP. It is not now possible to observe operations in 2015, and there is a clear risk of misinterpreting factored data, but based on the sources available, logically either the service’s “Intercity” seats between Catalunya and Navarra are extremely limited, perhaps to 10 or 20 each day – which judging by their frequent discounts, seems entirely unlikely – or these long distance “commercial” Intercity fares are effectively being subsidised on the back of a local OSP service.
Aside from casting further doubt over prenouncements of Renfe’s commercial profitability on longer-distance passenger services, any internal cross-subsidisation is largely irrelevant while Renfe continues to operate a state monopoly. Any future conflation of OSP and commercial service delivery could certainly pose problems under liberalisation (from 2020), especially given the European Court of Justice’s clear rejection of the cross-subsidisation with the potential to distort competition. But any market entrant expecting the terms of the competition to be fair has already fatally misread the field: What is fair will surely be what is fair for Spain, and as discussed in Arriva Celta, the challenge to new entrants is to create a liberalised market, not just exploit one.
From Extremadura to Catalunya Nord
If the cost-saving route rationalisations of the mid-1980s marked a low point in RENFE‘s prowess, Alta Velocidad surely heralded the operator’s rebirth. But as France eventually recognised, AV‘s most tangible costs transpire to be that which is not spent on the rest of the transport network, what this essay characterised as cascade effects: The implicit, and increasingly explicit, need to improve every lesser class of transport to maintain balance against the inflated public expectation that AV fosters. Fortunately Renfe had always had to manage such inequality, and apparently did so by compensating passengers on lesser, specifically slower, trains with a lower fare – monetary compensation for the inadequate performance of the state. Unfortunately AV was never placed in this context, presumably because it was initially defined by aviation markets – but from the moment Ciudad Real (a city which subsequently proved itself not part of that aviation market) was added to LAV, AV became intrinsically linked to Renfe’s traditional, often relatively local, passenger railway functions. Having subsidised the capital costs of AV, such that AV could match aviation on price (rather than taxing aviation and letting AV fund itself), rebalancing AV against non-AV implied either far greater non-AV fare subvention, or the improvement of non-AV services – neither of which occurred sufficiently to affect a balance.
In the 1990s AV was new, and even Barcelona could be convinced by Josep Borrell to wait in line. In the 2000s Aznar’s LAV network was somehow under construction just about everywhere. In the 2010s the Crisis was to blame for stalling construction just about everywhere. But as the 2020s dawn, and AV is somehow still an “ilusión” in many of Spain’s smaller cities, the inequalities – not just between AV and none, but between AV and non-AV – are becoming plain to see. Spain is full of inequalities, which is why the inequalities have to be hidden, else state becomes unstable: Inequalities are not a problem, but perceiving them is. Extremadura provides a telling contemporary case study:
With an average density of just 26 people per square kilometre, its largest town populated by only 150 thousand, Extremadura is far from ideal territory for passenger railway transport. In 2015 Extremadura’s passenger trains were all OSP, thus INECO’s data can be crudely apportioned to Extremadura using the earlier method applied to train-kilometres: Roughly half a million train passengers each year, which is under 10% of all local scheduled public transport, or less than 1% of all passenger journeys in the region. Each rail journey is subsidised by about €20, twice the fare paid – an above-average subsidy of a distinctly average operating cost, for regional trains that travel at an above-average speed. Yet the perception of poor service, both scheduled and operational, combined with the unfulfilled promise of LAV, has turned the most locally irrelevant of transport modes into a rallying cry for the Community. In the wake of 2017’s protest march, the government of Mariano Rajoy was forced to yield a daily Talgo train in lieu of the previous Media Distancia service, albeit on a route that only covered half its costs from revenue (which perhaps explains why the new Talgo service is shared with Media Distancia over three segments that represent its entire journey – Badajoz-Cáceres, Cáceres-Talavera, Talavera-Madrid). And after a train fire lead to a humiliating walk for passengers across the summer fields of La Mancha, Extremadura’s president was dispatched to Madrid to bemoan the lack of “solidarity” of Spain with his Community. Naturally he returned with promises of infrastructure – track upgrades and maintenance facilities now, with LAV later – physical infrastructure being the basis of the political state and thus the only form of discourse. Yet the issue in Extremadura is not fundamentally about a failure to manage railways, but a failure to manage equality.
The modern Spanish political state is surely digging itself into a hole, inevitably unable to offer AV everywhere, and – as will surely be the case in sparsely populated Extremadura – increasingly required to support AV’s higher operational costs, not just LAV‘s initial construction costs. As demonstrated by its traditional regional fare structures, RENFE was conceptually better able to manage inequalities behaviourally – Spain as the behavioural patterns of her people, not just Spain as its territorial infrastructure. Yet modern Renfe evidently struggles to balance inequalities around fares, with even its more local operations now hamstrung by the cost structure of a remote aviation market. As AV gradually becomes the default speed for most longer-distance trains in Spain, Renfe is having to find increasingly ingenious ways to maintain non-AV long distance services, further blurring the boundary between national and regional for slower services, as seen on the Barcelona-Pamplona Intercity: The operational conflation of OSP and non, priced as a yield-managed regional fare.
Amigos Para Siempre? As a nation that doesn’t exist, Spain places a remarkable emphasis upon its transport infrastructure. Such a complex and incendiary statement frames Spain’s transport geography: The state manifests through physical infrastructure, here the transport network that physically binds Spain’s regions to one another as a nation. Yet the society that state represents is vastly more virtualised, albeit within discrete knowable groups which are insufficient to manage a whole nation as that whole. On the Wings of Hope introduced the idea that the relentless tension between transport as “presence” (state structure) and transport as “utility” (actual use), is driven by the (hopeful) acceptance of external global finance, which bolsters the internal state against the very external that feeds it – largely unaware that this external finance is increasingly providing the glue between discrete societal groups, a cohesion that was once provided by traditional counter-balances such as public competition.
Since the 1990s external finance has gradually infested Spain’s regional Autonomous Communities, yet not directly the very local town or city administrations within, whose financing retains a much stronger relationship to their respective local economies. That pattern points to a narrowing of knowable, familiar geography, such that the locality (town or city) is now the largest scale at which the geography is itself important. Catalunya’s Procés serves to demonstrate the ease with which two utterly polarised perceptions of the same region can emerge, and to a degree coexist – any conflict tokenised, as seen in the battle of the “yellow ties” that occasionally spills onto the beaches of the Costa Brava. In so far as it created a local variant of Spain – an idea of a place that varies by perspective – Catalan Independentism surely succeeded. And having created a local variant of Spain, Catalunya presumably now requires much the same emphasis upon transport infrastructure to bind it. The corollary, that prior differences in approach between national and regional transport have lost their importance, with the defining transition now between the national/regional and the locality.
The vast distortions between those tiers of government with access to external globalised finance, and those without, defines the type of transport system each tier can provide. The inter-dependency between these tiers (the city operating a transport system that only the region can raise sufficient capital to invest in) complicates contemporary structures, but in broad principle, those transport systems which are externally financed are those transport systems which are not serving a knowable, familiar geography. Consequently there should be a strong correlation between those transport systems which rely on external globalised finance, and those transport systems for which the use of external globalised business models, such as yield management, are culturally acceptable. The utopia, in both its modern literary meaning of “imagined perfection” and its Classical meaning of “non-place”, described in the first section – of transport managed by algorithms which users trust blindly to optimise their unfamiliar travel behaviour – neatly fits the Spanish societal model, which routinely trusts in the functioning of unknowably complex systems – or is inclined to demand them by “right” should those systems (in the widest sense, state) fail.
Perhaps what actually binds Spain is a pattern of behaviour, not a physical geography. The latter is an artefact of a theological root that still manifests the common political state physically – unable to respond to the growing importance of external finance in binding Spain’s discrete societal groups together – an inability to respond that was previously explained as the “AVE hypothesis” (the rationale for the continued construction of new high speed railway infrastructure). The political state is by definition a function of how its citizens perceive it, meaning the physical infrastructure remains an important part of how those citizens perceive the transport system, even where every other aspect of the societal and financial structure suits the external models, such as yield management. The combination implies a resolution based on the distinction between capital investment and revenue operations, where external finance delivers the expected physical infrastructure (the “presence”), and operations adopt an external yield management model (the “utility”). Such a division is nonsensical strategic business planning, since it splits investment and operational decisions that are functionally inter-related, but is indicative of a balance that would satisfy each individual requirement.
The counter-intuitive answer to the question posed by this essay’s title, “Is Alta Velocidad Fast?”, is no: LAV is built on external globalised finance, therefore matches external globalised business models, those where passengers pay not specifically to go fast, but rather pay to use fixed assets efficiently. The logic is that the transport geography in which external finance is applied equates directly to an unfamiliar geography, which is not a geography that needs to be kept stable. Or philosophically, in those areas where a society embraces the external global, it by definition cannot remain internal to itself. The political structure of state cannot manage this because it is rooted physically, but the society can, at least when transport is not framed as infrastructure. Herein a logical niche for a liberalised passenger railway operation – to serve the society and not the political state.
It is perhaps no coincidence that Renfe enters another turbulent period of its history with a Catalan at the helm, and not just because of the added political complexity of Renfe’s contract in Catalunya. Mid-distance services in the region are among those closest to profitability, among those most likely to challenge any formal division between public and private. For example, the Barcelona-Figueres Avant operates at a profit, even with a large proportion of its busy peak morning services missing from INECO’s data, but that profitability is premised on the low operating costs achieved by reserving seats on shared AVE trains – trains with a complex commercial status. Yet ultimately the more complicated question will be why that Avant service terminates at Figueres, and does not continue on to Perpinyà and Rosselló – areas with strong cultural associations to Catalunya, albeit also rather French, commonly called Catalunya Nord.
Iberian cross-border European Union railway policy has focused on Trans European Networks, promoting the ludicrously outdated notion that AV can compete with aviation over distances of the magnitude of Paris-Madrid. TENs‘ nation-to-nation structure clearly suits both French and Spanish state centralism and the natural foci of their respective national railway companies, but is inclined to disregard the mid-distance markets where AV could be competitive, and has done nothing to integrate the local services that cross borders. But if, as concluded by this essay, the transport geographies of region and nation are now indistinguishable, with only locality requiring protection from the external world, an Avant from Barcelona to Perpinyà is just as reasonable a proposition as that to Figueres.
As discussed in Arriva Celta, such international links may be operated under the third package of European railway policy, not the fourth due in 2020, so pose a pre-liberalisation test. And as described in that essay, test is the operative word: Air Nostrum’s ILSA – a business that still lacks trains and drivers – is evidently under-prepared for actual operations between Madrid and Montpellier on the 5th October. ILSA’s proposal is ostensibly the notification of a challenge to Renfe’s patrimony. Albeit one that also grants ILSA the only formally-officiated permission to operate the cross-border AV service that the Generalitat de Catalunya, the region’s autonomous government, has been musing upon for years. The cross-Pyrenean Perthus Tunnel may finally see the traffic it was forecast, albeit not in time to have saved the tunnel’s concessionaire from bankruptcy.
While easily portrayed as a recalcitrant artefact of a bygone era, Renfe transpires to be a better manager of state than the state. That’s surely why Renfe, in its various guises that start with Explotación de Ferrocarriles por el Estado in 1926, has survived every fluctuation in government and policy – of which there have been many – all while lovingly tending the line to Ferrol that no concessionaire ever wanted. Transport economists and European technocrats are just the latest in a sequence of well-meaning Inquisitors, whose infallible liturgies inevitably fail to comprehend the myriad of discrete relationships and counter-balancing decision-making processes that manage to maintain a national railway in a nation that – in so far as its society is highly virtualised – doesn’t exist. Of course another operator in another nation would do things differently, not least tying themselves in less contorted numerical knots while trying to justify centrally what happens locally, as if hierarchy is the norm. But another nation contains another society, and comparison is not always fair or welcome.
There is a fear, particularly on the left wing of Spanish politics, that the behemoth of global capitalism, intent on optimising the world for analytic efficiency, will destroy the familiar localities (societal, organisational, geographical) that people know. As this and the Art of Public Competition essays have demonstrated, that threat is only as strong as its reliance on that global capitalism, specifically here external finance. In practice the greater threat is that posed to the territorially-focused right, whose physical nation-building cannot so easily be detached from external sources of finance. Spain’s regions transpire to now be no more knowable than the nation, “Catalunya” providing the extreme example, and thus transport geographies of a regional character should require no greater protection that those of the nation. The boundary between nation/region and locality (town/city) will inevitably be a blur, but can be defined by its degree of reliance on external finance. The principle is simple – that dependence on the external world means becoming part of that world. The flaw in the contemporary Spanish ilusión is surely the presumption one can enjoy the fruits of the global world, while maintaining the comfort of one’s familiar environment.
Alta Velocidad is not fast. AV‘s transport geography is not familiar, so it does not require the consistency of a fixed fare structure, hence can use external business models such as yield management, and should do so because those correlate with AV’s dependence of external capital finance. The statistical outliers to that conclusion, Avant’s daily commuters, are in the meantime the most exposed element of AV’s unresolved inequalities. Renfe’s regional fare structure compensates passengers that travel slowly, providing a behavioural means of balancing inequalities across its non-AV network. But having matched, rather than taxed, aviation, the cascade effects (of inflated expectation) to lesser classes of transport can only be addressed through subsidy in their operations, or as investment. Here, and in the more widespread adoption of yield management, the biggest stumbling block is the adherence of the modern Spanish political state to physical infrastructure. The inability to deploy behavioural policies, including those to which Spain’s society is naturally attuned, greatly inhibits the government’s ability to manage inequality. That the society is better at managing this than the state, will ultimately find its own balance – and perhaps more easily with the reduced state involvement sought by liberalisation. The corollary, that liberalisation liberates Renfe.