Research in Transportation Economics 100 (2023) 101310 Contents lists available at ScienceDirect Research in Transportation Economics journal homepage: www.elsevier.com/locate/retrec Research paper The impact of ‘competition for the market’ regulatory designs on intercity bus prices Javier Asensio a, b, *, Anna Matas a, b a Department of Applied Economics Universitat Autònoma de Barcelona (UAB), 08193, Bellaterra, Barcelona, Spain b Institut d’Economia de Barcelona (IEB), c/ John Maynard Keynes 1-11, 08034, Barcelona, Spain A R T I C L E I N F O A B S T R A C T JEL classification: Spain regulates its intercity bus market by means of a ‘competition for the market’ mechanism, whose design has D44 been modified several times in the last years. This implies that current services are operated under contracts L51 whose conditions are heterogeneous. We take advantage of such fact to empirically measure the impact that L92 regulatory designs may have on fares paid by the users. Controlling for the different determinant of bus prices at R48 route level the results show very large differences between routes whose contracts were awarded under relatively Keywords: open conditions compared to regionally regulated routes or old contracts whose concessions were extended in Competitive tendering Intercity buses 1987 and have not been retendered since then. The observed difference between the cheapest and the most Competition for the market expensive services is to a great extent explained by the difference in the regulatory designs used to award each Regulation contract. 1. Introduction participate), or promoted more aggressive price offers by participating firms. Different European countries have decided to liberalise their inter­ In practice, substantial price differences per unit of distance are city bus markets. Following the UK experience in the 1980s, countries observed in different routes. Taking an extreme case, in early 2021 a such as Sweden (1998), Norway (2003), Germany (2013), Italy (2014), customer travelling between Madrid and Segovia (distance 97 kms) France (2015) or Portugal (2019) have allowed for the entry of new would pay a fare of 4.16 €, or slightly more than 4 cents per kilometre, operators who can compete with each other or the incumbents in the while a trip between Barcelona and Lleida (148 kms) had a price of market.1 In Spain, on the contrary, such entry is strictly limited, and 22.27 €, or 15 cents per kilometre. Such differences may be due to very competition takes place ‘for the market’, as firms need to bid in order to diverse factors, such as existence of alternative modes of transport, win a contract with the public sector which specifies the services to be consumers’ income levels or operators’ costs. However, they may also provided. respond to differences in the regulatory framework and the contract However, there are substantial regulatory differences among routes characteristics under which each service is provided. within Spain. Not only are intra- and inter-regional services subject to The changes in the terms of tender offer a good scenario in which to different norms, but the latter’s regulations have experienced frequent test up to what point incentives to compete for the contract may have an changes in the recent past. This implies that the conditions under which impact. In particular, the aim of this paper is to empirically test if current services are operated may be very different, as they depend on observed price differences can be due to the differences in the regulatory the regulatory regimes under which each contract was awarded. One regime under which each service operates. route may have been awarded under regulatory rules that gave very Little is known about the specific impacts that contracting out little importance to prices, or which favoured the incumbent operator, characteristics have on the bids of firms competing in the bus tendering while other regulatory regimes may have provided more incentives to market. Some papers have analysed how the design of the terms of bidding by firms (such as supressing unnecessary requirements to tender may affect the operating costs or efficiency of operators, with the * Corresponding author. E-mail addresses: javier.asensio@uab.es (J. Asensio), anna.matas@uab.cat (A. Matas). 1 See Aarhaug et al. (2018), Aarhaug and Fearnley (2016), Alexandersson et al. (2010), Beria and Bertolin (2019), Beria et al. (2018), Blayac and Bougette (2017), Dürr and Hüschelrath (2015), Reynolds and White (2022), Thust et al. (2016) and van de Velde (2014) for different analyses of the liberalisation process in these countries. https://doi.org/10.1016/j.retrec.2023.101310 Received 21 November 2022; Received in revised form 12 May 2023; Accepted 28 May 2023 Available online 23 June 2023 0739-8859/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). J. Asensio and A. Matas Research in Transportation Economics 100 (2023) 101310 assumption that such cost decreases would lead to better outcomes for included in a particular version of the terms of tender as a regulatory consumers. Based on competitive tendering of local buses in Germany, design. Usually, a given regulatory design is used for a set of contracts Beck and Walter (2013) show that setting appropriate tender conditions awarded during a period, with the Ministry of Transport declaring that regarding vehicle schedules, frequency and timetables increases the its purpose is to keep them constant for as long as possible. However, as operators’ ability to optimise circulation plans and schedules, with a will be mentioned later, in recent years there have been frequent consequent cost-decreasing impact and, hence, lower prices. Conversely, changes in the regulatory design. We exploit such variability to identify introducing vehicle requirements that increase the risk borne by the the impact on the prices paid by the users. operators has the opposite effect. Using aggregate Swedish public In the case of inter-regional services, a contract typically combines transport data Holmgren (2013) estimates a decline in the average ef­ high-density routes that can return positive margins together with ser­ ficiency of operators over time. Among the factors that can explain it, vices connecting with rural or low density areas, which are usually loss- Holmgren points at increases in network density and detailed demands making ones. As services are provided as net-cost contracts without any on vehicles’ design. Nonetheless, such requirements can be partially operation subsidies, the system implies a cross-subsidisation from users related to stricter environmental and safety standards, which is not of profitable services towards those in routes that run at a loss. However, necessarily bad. Vigren (2018) finds that different forms of contract the operator of a loss-making route can apply to the regulator for a designs have an impact on the number of bidders participating in ten­ modification of specific itineraries or frequencies, or even for a merger ders for bus contracts in Sweden. Given that for tendered bus contracts of different contracts provided by the same operator. Intra-regional more bidders are viewed as a way to improve tendering results (Can­ routes very frequently receive subsidies directly from the regional gov­ tillon and Pesendorfer, 2006), tender characteristics that increase bid­ ernments, under Public Service Obligations schemes. ding participation may reduce prices. Along the same lines, using data of At the time of passing the LOTT, in 1987, the then existing services London bus contracts, Amaral et al. (2013) find that a higher number of were automatically turned into proper concessions, directly awarded to bidders is associated with lower costs of service. Our paper contributes its incumbent operator and given a 20-year extension of the contract. to this literature by providing evidence on how specifications and re­ Other services were created as new routes, awarded by means of a quirements in contracts affect the prices offered by bidding operators. competitive auction under the rules previously described. These auc­ In order to do so, the next section explains the main features and tions were very contested, with up to 29 firms bidding for the Madrid- recent evolution of intercity bus regulation in Spain. Section 3 reviews Málaga-Algeciras contract or 23 for the Irún-Algeciras one. Such level of the recent literature on the estimation of price equations for long- competition for the market resulted in lower prices than those offered by distance bus services. Section 4 describes the dataset collected for this the services whose contracts were extended (OECD, 2001). For instance, empirical research, while section 5 presents the results of the econo­ prices on the Madrid-Sevilla route, which was tendered at that time, metric model we use to test our main hypothesis. Section 6 discusses our were 30% below those of the comparable Madrid-Bilbao one, whose findings in the light of the design and effects of contracting out bus concession was one of the extended ones (Barrio, 2013). services. Section 7 concludes. The initially expected period of 20 years during which no contracts would be re-tendered was extended in some cases by means of different 2. The Spanish intercity bus market measures whereby operators would be granted longer operation periods if they kept price increases under certain limits. This policy can be un­ In Spain, all regular intercity bus services are provided by private derstood in the context of Spain’s efforts to meet the Maastricht Treaty operators acting as regulated monopolists during a given period of convergence criteria in the second half of the 1990s, and in particular time.2 Two types of situations can be identified, depending on the inter- the one referring to inflation control. Other extensions were granted as a or intra-regional character of the services. In the case of services that consequence of particular mergers between contracts operated by the connect origins and destinations in different regions of the country, the same or different firms. regulatory authority is the national government’s Ministry of Transport. The result of such extensions was that, contrary to what would be For those routes that take place completely within a given region, the expected given the 20-year limit set in 1987, not all the re-tendering Transport Department of the corresponding regional government will be processes took place in 2007. On that date, the Ministry of Transport responsible for awarding and regulating the provision of the services. reached a compromise with the business association of bus operators The main piece of legislation is the 1987 Transport Law (LOTT),3 running the services and the trade unions representing their workers, by which established the current system of services’ tendering by means of means of which the terms of tender to be used in the retendering of a multidimensional auction mechanism. This implies that a pre-specified contracts were agreed on. These terms of tender were severely criticised score function is used to compute the points obtained by each bidder on by Spain’s competition authority on several grounds (CNC, 2008). each variable, weighting the scores to result in an outcome on a 0–100 Among the most important issues were the low weight given to price and scale, with the winner being awarded the contract. frequency bids in the score function together with the upper and lower The decision as to which variables are included in the score function limits set for such bids, the requirement of already being an operator of a and the weights given to each one of them are taken by the regulator regular service to submit an offer, the broad discretion in the assessment when designing the terms of tender for each auction. The law allows the of non-economic criteria and the right of preference for the incumbent in regulator substantial leeway in such choices, although the price paid by case of tie. As a result, the Ministry modified the terms of tender to be the user and the level of frequencies4 always need to be included as used in 2009. However, leaving aside a slight increase in the weights variables in the score function. We refer to the variables and weights given to price and frequency and some other minor changes, the main anticompetitive terms remained (CNC, 2010). By that time, some bus companies not operating regular services had taken all the terms of tender used since 2007 to the Administrative 2 Although the time length of the contract is clearly specified in the terms of Court, where a ruling declared them void. Thus, the Ministry of Trans­ tender, delays in the re-tendering procedures may result in longer periods than port had to issue a new version of the terms in 2011 which could be initially planned. In early 2022 more than 50% of the contracts were beyond considered to be more pro-competitive. Among the favourable changes their expected re-tendering timetable. 3 were a new score function that gave 40% of the weight to prices and Ley 16/1987 de Ordenación de los Transportes Terrestres (LOTT). 4 Operators bid in terms of ‘bus-kms’ offered, for a given itinerary, implicitly frequencies (25% and 15%, respectively), the reduction of the duration defining the level of service frequency. Here, we interpret the term ‘frequency’ of the contract and the easing of the conditions to participate. However, as services per unit of time, and not as the time gap between services, so that those terms of tender were also contentious since they made it higher frequencies make the service more attractive to the users. compulsory for the company winning the contract to take over the 2 J. Asensio and A. Matas Research in Transportation Economics 100 (2023) 101310 incumbent’s operator labour force assigned to that particular contract, to less pro-competitive conditions. Among the changes, it is worth under their existing labour conditions.5 Other remaining features of the noting a decrease in the weight given to price and frequency bids to 51% terms of tender that limited competition were the scoring system for and a new score system for the price variable that severely limited the price bids which the facto kept a lower limit that prevented full point difference awarded to below average bids. Additionally, a new competition on prices and the maintenance of the right of preference for ‘podium’ system was introduced to value some of the items requiring a the incumbent in case of tie. qualitative assessment in such a way that only the best bid obtained the Only five contracts were tendered under those terms before the maximum number of points, the second got a lower score and from the regulator was again taken to court by some firms, who won their case. third onwards the number of points was again reduced. In comparison Therefore, by 2014 another version of the terms of tender was issued with the previous system, according to which each bidder could obtain which finally suppressed those features that were judged to be anti­ the maximum number of points in each section, the podium system may competitive by the courts (CNMC, 2014). By that time, a significant introduce distortions. For instance, when technical offers are very delay had built up in retendering expired contracts. Regarding the score similar the system widens the differences, and if an offer is well above function, the new system established a clear distinction between criteria the average the system does not reward it proportionally. that are assessed according to value of judgements and those that rely on These terms of tender were used for just two contracts held in 2018, a formula. The first group includes variables such as vehicle comfort, before a further modification took place in 2019 with the objective of customer service, measures addressed to obtain security certificates, restoring a more-competitive environment for bus tenders. Among other energy efficiency programs and incentives to connect with other public changes, the maximum duration of the contract could be set below 10 transport modes.6 In the second group fares and frequencies are the most years, the obligation of the contract winner to pay a 1% revenue annual important variables. Additional variables that were also assessed by fee was supressed and the score system was similar to that of 2016 means of a formula are vehicle age and the commitment to fulfil quality tenders. Only two contracts were issued using this system. Besides, in all standards according to pre-established regulations. The 2014 terms of tenders issued in 2018 and 2019 the regulator suspended the procedure tender gave a 55% weight to price and frequency variables, which were before the contracts were finally awarded, so we observe no price data in assessed with a proportional score system in order to encourage routes subject to such regulatory designs and, consequently, none of competition. However, upper price and lower frequency limits were still them enter our sample. imposed, and in 2015 the Ministry of Transport applied significant in­ Table 1 provides a summary of the variables entering the score creases in the maximum bidding prices and reductions on the minimum function together with the corresponding weights. When reading this frequency limits without any explanation as to how the magnitude of table, two facts must be taken into account. Firstly, changes in the def­ those changes was decided. Those terms of tender also extended the initions of variables make it difficult to classify all items entering the compulsory taking-over of the incumbent’s labour force to all workers in score function into the same broad categories. Hence, comparisons related activities, such as fare collection, maintenance, or administrative should be made with caution. Secondly, besides the score function, the duties. Given the ability of the incumbent to flexibly define such as­ terms of tender included a set of compulsory requirements regarding signments, this requirement can be regarded as an important advantage vehicles, labour force take-over and the fulfilment of quality regulations for the incumbent, as it would be better able to integrate its own together with guarantees of technical, financial and professional ca­ workforce. Other features that could be detrimental for competition pacity of the firm. Therefore, the technical issues included in the score were a fee payable to the Ministry of Transport equal to 1% of the function have to be interpreted as improvements above the standards winner’s turnover and a participation requirement consisting in having established in the contract. at least three years of experience in equivalent services with a bus fleet In summary, from 2007 onwards (with the exception of 2018), bid­ of similar size. ding for long distance bus services was developed in an increasingly pro- In 2016 a new version of the terms of tender was issued that, without competitive environment. The results of the different terms of tender essentially modifying the requirements to bid and the structure of the show that score differences in the technical part – evaluated according to score function, increased the importance given to fares up to 45%, while value of judgements – have been limited since all bidding companies can that of frequencies fell to 10%. Additionally, the computation of both have the latest bus models, with similar characteristics for comfort, se­ price and frequency bids was modified from the previous proportional curity and energy efficiency and can also access the same technologies to system to a new two-part method, with different ways of computing the provide services such on-board internet connections. Hence, competi­ points depending on whether the bid was on the upper or lower half of tion has taken place mainly on price and frequencies. Despite the scores the distribution of submitted bids. The system was designed so as to given to prices and frequencies, there is no evidence of price competition encourage more aggressive price and frequency bids.7 at the expense of frequency of service. In more than 50% of cases the These terms of tender raised strong criticism from the incumbent winner bid presented the highest bid on frequency. Nonetheless, a large operators, to which the Ministry of Transport was sensitive. In 2018, number of suspensions in the procedures due to Court rulings or appeals new specifications for the tenders were approved which meant a return makes it difficult to assess the actual results. In addition to inter-regional services, our sample also includes intra- regional ones whose regulation, as has already been mentioned, depends 5 on the regional governments. Some regions have passed their own The previous terms of tender had given 20% of the total score to firms transport laws, which may allow for explicit subsidies for loss-making accepting such compromise, making it de facto a standard practice, since no contract was winnable without it. services, while others follow the LOTT procedures. Nonetheless, in all 6 Specifically, some of the variables related to vehicle comfort are vehicle of them the tendering systems prevails. Intra-regional services may have ergonomics or internet connection, whereas those related to attention to public some distinctive characteristics as joint allocation of urban and regional include compensation for delays, complementary insurances and loyalty pro­ services, on-demand services in rural areas, joint provision of schooling grams for frequent users. and regular services or subsidised contracts of different intensity. The 7 Specifically, for prices (frequency) bids above (below) the average, the available information does not make it possible to characterise each points obtained varied with the bid at a constant rate. However, for bids below route according to the terms of tender designed by each regional gov­ (above) the average the points obtained increased more than proportionally as ernment, as we have done for the inter-regional ones. Therefore, in the the bid decreased (increased). The exact formula of the points obtained is given empirical analysis that we report below, we group all regional routes by 45⋅PPmin bid , for bids lower than the average price bid (Pavg), where Pmin is the into a single category. It may well be that some of the specific charac­ lowest price bid (which would obtain 45 points). However, for bids above the teristics of such contracts increase the operating cost and, consequently, (Pmax − Pbid ) average the formula is 45⋅PPmin avg (Pmax − Pavg ) . distort the comparison among regulatory regimes. However, given that 3 J. Asensio and A. Matas Research in Transportation Economics 100 (2023) 101310 Table 1 Weights given to bid variables (%). Terms of Price Frequency Labour force take over Security and Customer Energy Public transport Other Total tender comfort service efficiency promotion 2007 10 5 24 26 13 8 – 14 100 2009 15 8 15 26 13 5 – 18 100 2011 25 15 compulsory 24 6.5 7 3.5 19 100 2014 35 20 compulsory 12 10.5 5 9.5 8 100 2016 45 10 compulsory 13.5 11.5 5 5 10 100 2018 35 12 compulsory 18 5 8 10 12 100 2019 49 20 compulsory 19 4 6 11 0 100 we only consider connections between provincial capitals our under­ is interacted with the country dummies, the impact is strongest in Spain, standing is that services are offered on similar grounds and are, there­ implying that market structure has larger impact in terms of higher fore, comparable.8 prices. The authors interpret this as a result of barriers to entry in Spain, Summing up, current routes are operated under contracts that may which is the only one of the sampled countries without a liberalized have been awarded under one of the following nine regulatory regimes: market. the concessions extended when the LOTT was passed in 1987; the new Beria and Bertolin (2019) study the Italian case, estimating a price contracts awarded competitively as the LOTT came into force; contracts equation for services operated during the year 2016. As all the previ­ which are the result of mergers between concessions of the two previous ously cited papers, they too find a negative impact of travel distance on types; the terms of tender designed in 2007, 2009, 2011, 2014 or 2016 prices. Besides, there are significant differences depending on the and, finally, intra-regional routes. operator providing the service. They also obtain results pointing at Our aim is to test whether such regimes have significant impacts on intense substitutability between bus services and railways or carpooling. the prices currently paid by users. In order to do so we estimate a price equation, where additionally to the usual determinants of prices, we will 4. Data and variables include dummy variables corresponding to the specific regulatory design under which the contract was awarded. In order to empirically test the hypothesis previously outlined, we have collected data on a sample of intercity bus routes in Spain. One of 3. Previous literature on price determinants of long-distance bus the problems of gathering representative data for this purpose is to services define routes where services provided are comparable in nature (i.e., not mixing rural or on-demand services with routes in high-density corri­ The estimation of price equations for bus services has become a dors). We therefore restrict our choice to connections between provin­ common tool to assess the consequences of the liberalisation, as shown cial capitals,9 which usually corresponds to the largest city in each by different papers in the transport economics literature. Augustin et al. province. This results in a geographically balanced sample of origins and (2014) estimate price equations for services in the United States and destinations. Therefore, connections between smaller villages, or be­ Germany, defining the dependent variable as the price per distance. As tween capitals and those villages, are not considered. This implies that explanatory variables they consider the journey’s distance and time, the sample cannot be regarded as representative of the whole market for number of stops and average service frequency, international connec­ regular bus services in Spain, as a substantial number of routes provide tions (in the United States) or services to airports (in Germany), as well connections to towns or villages with smaller populations than the cities as a set of measures to take into account intermodal competition (train sampled here. However, it should be a valid sample for our purpose of price, journey time and need to change trains). To observe the impact of measuring the impact of regulatory changes on prices paid by bus users. competition, their US model includes the number of operators in the It should also be noted that although the services observed in our route, while dummies for the largest firms are considered in both cases. data are direct, in the sense of not requiring changing buses, they are not The reported results show that bus prices decrease with the journey’s necessarily non-stop express connections, as there may be some stops distance, although the inclusion of the number of stops as an additional along the journey. A connection in our sample is typically part of a variable makes it difficult to interpret the impact. In the US, prices longer bus route that links different cities. decrease with the number of competitors, while the firm dummies reveal We collected prices for regular services for travel on a particular date significant price differences between Greyhound and Megabus. In the (September 23, 2019), three days prior to the trip. Prices were collected case of Germany, the dummy for Flixbus services obtains a negative and from checkmybus.es, a website providing prices and timetables for each highly significant coefficient. Also for Germany, De Haas, Herdold, & bus service, but not directly selling tickets. In the cases where a Schäfer, 2022 report that the takeover of PostBus by Flixbus (which connection is offered by more than one available direct service, we coincided with the exit from the market by BerlinLinienBus), had the compute the average price. Additionally to the price and frequencies impact of increasing fares by between 8.3% and 9.3%. (number of services), other variables have been obtained from different Fageda and Sansano (2018) estimate price equations for intercity bus sources. From official timetables we are able to obtain both road dis­ services, collecting data for a sample of national routes in Spain, France, tances between each origin and destination as well as travel times, Italy, Germany, Sweden and the United Kingdom. They report negative making it feasible to compute average speeds in each service10. We impacts of route distance and speed on prices per kilometre. They also would expect a negative sign of speed in the price equation, as it reveals find that the presence of alternative rail services diminishes prices, while that of air connections does not. The coefficient of their national market concentration measure (HHI index) is not significant. However, when it 9 Spain is administratively divided into 17 ‘Autonomous Communities’, to which we refer as regions, and a total of 50 provinces within them. Only the 47 peninsular provinces are considered here, as the 3 which are part of the Balearic 8 In order to check if our results are sensitive to the inclusion of the regional or Canary Islands have no long-distance bus services. 10 routes, we estimate our econometric model excluding those observations. The For intra-regional services we do not have access to official timetables results are reported in Table A2 in the Appendix and are shown to be very showing exact route length. We obtain such distances as the best option for car similar, so our main conclusions do not vary. travel from available map sites. 4 J. Asensio and A. Matas Research in Transportation Economics 100 (2023) 101310 Table 2 synthetic index13 of the overall attractiveness of each contract, we first Price equation variables. Descriptive statistics. compute for each route the ratio of the average population of the mu­ Obs. Mean Std. Dev. Min. Max. nicipalities it covers over the product of the number of stops times the total distance of the route. The idea is that a short route without stops Price (€/km) 221 0.076 0.017 0.042 0.150 Distance (km) 221 490.3 271.9 60 1255 connecting cities with large populations will be more attractive that a Income pc (€) 221 24250 4412 20512 36921 long one with multiple stops in small villages. Then, the attractiveness Population (000) 221 868.2 953.7 125.3 3719 index for the whole contract is obtained as the weighted average of the Plane (dummy) 221 0.06 0.24 0.00 1.00 indexes computed for each one of its routes, where the weights are the Speed (km/h) 221 74.4 9.7 46.5 101.3 Bus/Rail travel time 221 1.54 0.61 0.34 3.21 number of residents living in the municipalities served by each route.14 Attractiveness index 221 144.52 265.09 5.29 2184.28 We expect a higher value of the index to be associated with a lower price. Table 2 shows the descriptive statistics of the different variables, while in Table 3 the average values are computed for the different types the ability of the firm to provide more journeys per bus in a given time of regulatory designs, as described in section 2. It can be observed that, period, implying that buses of a smaller average size and fewer drivers without controlling for any other variable, prices are higher for intra- may be needed (Fageda and Sansano, 2018). regional and LOTT routes. However, in order to properly assess the Distance is used to build a relative measure of our dependent vari­ impact of regulatory designs we need to estimate an econometric model able (price per km), but it is also included as an explanatory variable in that includes all potential determinants of prices. order to test whether costs may vary with the route’s length in the provision of bus services and the passing of its generated savings to 5. Estimation and results consumers. We define the market size for each route as the sum of the population The price equation that we estimate can be represented as of the two capitals it connects. The impact of market size on prices can ∑R be the result of two opposite forces. On the one hand, larger markets will ln pkmi = α + βXi + γ dri + εi (1) r=1 r result in more demand for bus services, which would push prices up­ wards. On the other hand, in competitive markets where demand levels Where pkm is the dependent variable (price per kilometre paid in 2019, are higher than the minimum efficient scale, market size can be expected in euro cents), the variables in matrix X include a set of routes’ char­ to be negatively correlated with prices, as the entry process of new firms acteristics: distance, income, speed, population, the ratio of travel times will typically push prices down. In our case, however, market entry is by bus over train and the contract’s attractiveness (all in natural loga­ strictly regulated, so that the only response to higher levels of demand rithm terms), as well as the plane dummy multiplied by the log of dis­ comes from the price bids that the operators may make in the auction, or tance. Besides, the d are dummy variables that take the value 1 if route i from the participation of a higher number of operators in such auction. is operated under a contract awarded according to the regulatory design In both cases, the role of competition should result in lower prices being r of the R categories listed at the end of section 2. The concessions paid in larger markets. extended by the LOTT are taken as the reference category. The term εi is We also include in the model the average of the two connected the error term assumed to be normally and independently distributed. provinces’ per capita incomes. Higher incomes should generate more We are interested in the impacts on prices revealed by the γ parameters. travel demand, either directly from final consumers or as a result of An important issue is whether changes in the regulatory design can higher levels of economic activity. However, intercity bus services be considered exogenous in the price equation. As explained in Section probably are an inferior good for many users,11 when other faster and 2, all the changes in the terms of tender have been forced either by the more expensive alternatives are available (train, private car or plane). competition commission or by the Administrative Court. Therefore, the Therefore, the coefficient of this variable could take any sign. usual concern for endogeneity due to reverse causation would not apply We consider the impact that the availability of alternative modes of in this context. transport may have. In the case of air services, we construct a dummy The results of estimating equation (1) by OLS are shown in Table 4.15 variable showing whether direct connections existed at the time be­ For variables expressed in logs, their coefficients can be directly inter­ tween airports in each province. Given that such services would be preted as elasticities. In accordance with the previous literature that unfeasible for very short routes, we interact that variable with the log of estimates price equations for bus services, the coefficient for distance is distance (so that the relationship is not necessarily linear). To consider found to be significantly negative. Prices per km decrease with the trip’s the competition between bus and rail services, we build a variable length. This result can be interpreted as a type of increasing returns in showing the relative travel times between origin and destination in each terms of distance in the supply of bus services which are, at least in part, mode.12 Slower bus than rail services would increase the value of such passed on to consumers. In the case of income, we obtain a negative variable. Therefore, we expect a negative sign for its coefficient as firms coefficient, consistent with the previously discussed characteristics of would need to compensate longer travel times with lower prices. intercity bus services as an inferior good. Market size, measured by Finally, we also consider an explanatory variable that measures how population sizes, has a significantly positive impact on prices, implying attractive a given contract may be for the private bus companies. As has that demand effects outweigh any potential price-reduction due to already been mentioned, routes with different potential profitabilities are grouped into a single contract, with implicit cross subsidies among them. Therefore, the prices set on a given route will not only be affected 13 See OECD (2008) and Greco et al (2019) for methodological discussions on by its own characteristics, but also by the relative attractiveness of the synthetic (or composite) indexes, and Groh et al. (2010) or Lieser and Groh whole set of routes with which it is grouped. In order to obtain a (2011) for applicaitons to capital and real estate markets, respectively. 14 This method is valid only for inter-regional routes, included in contracts awarded by the national government. For intra-regional connections we compute an equivalent index on the basis of the non-stop and multiple-stop 11 If intercity bus services are an inferior good, increasing incomes result in a services in each route, assuming that both types of services constitute a leftward demand shift, thus decreasing both equilibrium price and quantity. contract. 15 Paulley et al. (2006) provide evidence of negative income elasticity for bus The results reported in Table 4 correspond to robust standard errors. The demand. results corresponding to clustered standard errors at contract level, which may 12 The methodology followed to obtain such measures is detailed in AIReF better control for possible autocorrelation between the routes in each contract, (2019). are reported in the appendix. 5 J. Asensio and A. Matas Research in Transportation Economics 100 (2023) 101310 Table 3 Variables’ mean values by regulatory design. Regulation Obs. Price (€/km) Distance (kms) Income pc (€) Population (‘000 inhabs.) Plane (1/0) Speed (km/h) Bus/rail time Attract. index LOTT-ext. 70 0.083 469.1 25 625 1250.5 0.08 74.68 1.62 215.2 LOTT-new 1 0.069 198.0 22 666 3340.8 0.00 72.00 3.17 307.8 LOTT-merged 46 0.074 491.1 23 593 813.1 0.11 71.63 1.46 111.9 2007 20 0.065 631.2 22 267 327.0 0.05 73.06 1.26 11.62 2009 10 0.056 871.2 25 650 1192.8 0.10 80.68 2.02 21.18 2011 16 0.061 478.1 24 688 540.2 0.00 80.51 1.31 84.07 2014 22 0.064 795.4 25 363 585.5 0.05 74.45 1.65 12.85 2016 2 0.043 183.1 24 834 2503.8 0.00 70.07 2.26 1214 Regional 34 0.092 173.2 23 498 546.8 0.00 73.60 1.44 735.7 Table 4 Estimation results. Coefficient Std. Err. t-stat Constant 0.428 0.864 0.50 Ln(distance) − 0.081** 0.029 − 2.81 Ln(income pc) − 0.156* 0.077 − 2.03 Ln(population) 0.103*** 0.015 6.84 Ln(speed) − 0.474*** 0.102 − 4.65 Ln(bus/rail time) − 0.099*** 0.027 − 3.62 Plane x ln(distance) − 0.012* 0.005 − 2.19 Ln(attractiveness) − 0.023 0.013 − 1.81 Tenders LOTT-merged − 0.131*** 0.027 − 4.84 LOTT-new − 0.307*** 0.043 − 7.09 2007 − 0.197*** 0.037 − 5.38 2009 − 0.314*** 0.036 − 8.63 2011 − 0.214*** 0.032 − 6.57 2014 − 0.239*** 0.053 − 4.51 2016 − 0.769*** 0.054 − 14.19 Regional 0.056 0.056 1.00 R2 0.6258 Observations 221 (robust standard errors). Legend: *p < .05; **p < .01; ***p < .001. Fig. 1. Coefficient plot of estimation results Reference category: LOTT extended contracts. increased competition in larger markets. The availability of airplane services for long trips acts as a substitute to bus services, although with very small effects: for the mean length trip in the sample (490 kms), the existence of an alternative plane service implies a 3.3% reduction in bus 6. Discussion fares. In the case of railway services, the faster they are the lower the price paid by bus users. A 10% improvement in rail travel times with The main conclusion from the results reported in the previous section respect to the bus from their mean value reduces bus prices by 1%. is that contract design changes can have substantial impacts on fares We now focus on the dummy coefficients with which we want to test paid by users of regulated bus services. With the available data we are the response of prices to changes in the regulatory design. The co­ not able to identify the channels through which such changes take place. efficients and their confidence intervals are plotted in Fig. 1. The They may be a combination of the limits set on different variables by the reference category are the services that were extended as the LOTT was design of the contracts, the attraction of a higher number of bidders and passed in 1987. We observe that these correspond to the most expensive the strategies put in practice by each potential operator to try to secure routes in our sample (together with the services awarded by regional each contract. Our results imply that increasing the weight given to the governments, whose difference is not statistically significant). The most relevant variables, such as prices and frequencies, results in lower contracts with the largest negative impact on prices are those awarded prices for users. under the terms of tender of 2016, when the weight of prices in the score The analysis explained in the previous sections touches on several function reached 45% of the total. This version of the terms of tender can issues related to the design and effects of contracting out bus services, as be considered as the one encouraging a most competitive bidding have been discussed in previous Thredbo workshops (see Mekert et al., behaviour by the firms. 2018, van de Velde and Alexandersson, 2020, or Preston and Walters, These estimation results imply that the regulatory design under 2020 for accounts of those discussions). which the contract was awarded can explain price differentials of up to One of the topics directly related to our analysis is that of the optimal more than 2 to 1. This would be the case if we compare the mean of the design of contracts in terms of the relative importance given to price and predicted price if all routes were provided as LOTT extensions (0.087 quality variables (understanding that frequencies can be a measure of €/km) with the ones predicted if services were run under the 2016 service quality). Ridderstedt et al. (2019) discussed this issue in the regulatory design (0.041 €/km). Swedish context and considered it preferable for the regulator to set 6 J. Asensio and A. Matas Research in Transportation Economics 100 (2023) 101310 quality levels and allow for competition in prices. Hensher et al. (2000) The results are obtained by estimating a price equation for services are also critical on the use of scoring rules that combine both variables as between provincial capitals, which are provided by contracts awarded they may have adverse selection effects. In the context analysed here, under different regulatory settings. Controlling for the main factors that the problem may be made more evident by the relatively large number may determine price differences, a statistically significant effect is found of additional qualitative elements that the terms of tender consider, for the different types of terms of tender. Specifically, the highest prices adding complexity to the assessment of non-price variables. However, are observed in intra-regional routes or in those inter-regional ones those elements are mainly related to bus operation characteristics that whose contracts have not been retendered since the 1987 Law. As an most firms should be able to satisfy in a relatively homogeneous way. increasing degree of competition has developed since 2007, a reduction One conclusion that can be drawn from the analysis is the apparent in the price paid is observed. This reduction increases as the terms of lack of maturity of the regulation of Spanish long distance bus regula­ tender allow for more competitive bids to increase the probability of tions. As shown in the analyses reported by van de Velde and Alex­ success. When we compare the extreme cases, our estimates predict a andersson (2020), a sign of such lack of maturity would be the frequent 53% reduction in prices per km if the services extended by the 1987 Law modification of the regulatory framework by the transport authority as a had adopted the terms of tender used in 2016, which are the most pro- result of pressure from different interest groups. Although the Spanish competitive ones. The main changes introduced at that date implied system of ‘competition for the market’ has been in practice for more than giving a higher weight to prices in the auction’s score function as well as 30 years without experiencing substantial legal changes, it has reached a allocating points on prices in a way that encouraged more aggressive situation of ‘de facto’ blockade due to the countervailing pressure of bids. opposing interests. On the one hand, pro-competition agents such as the The analysis reported in this paper shows that under a ‘competition CNMC or the firms wishing to enter the market have had some success in for the market’ system the role of the regulator as designer of the obtaining increasingly open terms of tender. However, this has met the institutional mechanism used to award the contracts can have very opposition of most of the incumbent operators. As the Ministry of significant impacts on the outcomes. Maintaining such mechanisms Transport has been unable to unlock this situation, the retendering of relatively constant over time can be a very complex task, as the regulator more than 50% of the current contracts is overdue. Instead of proposing will receive pressures from different parties, including incumbent op­ further changes to the terms of tender, the Ministry of Transport plans to erators (who will not participate in the tenders if they do not consider aggregate the current 88 contracts into 22 much larger ones. If put in them attractive enough), consumers (asking for more affordable and, practice, this reform would provide clear evidence about the trade-off simultaneously, more frequent services) or even the government, who between benefits derived from economies of scale and the effects of may prefer a system based on internal cross-subsidies to one in which limiting competition by restricting participation in the market to large loss-making services are financed from the public budget. companies. At the time of writing (May 2023) a new Transport Law currently CRediT authorship contribution statement under Parliamentary discussion would allow for the opening to ‘competition in the market’ of some routes. This can be understood as an Javier Asensio: Conceptualization, Methodology, Software, Vali­ example of some movement along the regulatory or contracting cycle dation, Formal analysis, Investigation, Writing – original draft, Writing – (Dementiev and Han, 2020), whereby the inefficiencies generated by review & editing. Anna Matas: Conceptualization, Methodology, Soft­ regulations may end up leading to a change that opens the market to free ware, Validation, Formal analysis, Investigation, Writing – original entry competition. However, it is not clear how the new legislation draft, Writing – review & editing. would cope with potential monopolisation of the services by private firms, or the way in which loss-making services in rural areas would be guaranteed. Declaration of competing interest 7. Conclusions The authors have no interests that could bias or influence the results of this research. Unlike other European countries, Spain has not completely liberal­ ized its long-distance bus market, opting instead to introduce competi­ Acknowledgements tion by tendering sets of routes into single contracts. As in all cases in which the mechanism selected to provide transport services is one of We want to thank participants at Thredbo 17 for their useful com­ ‘competition for the market’, the design of the terms of the tender be­ ments and discussion, Peter White for commenting on a previous version comes a crucial aspect to define the prices and qualities provided to the of the paper and the referees for very useful comments. We thank Javier final users. Taking advantage of different regulatory conditions applied Gutiérrez and Juan Carlos García for sharing their computations of rail to inter-regional and intra-regional routes, as well as of changes in the travel time data and Ramon Bonvehí for excellent research assistance. details of the terms of tender for the former, this paper provides evidence This research has received financial support from BBVA Foundation and of the relevance of such designs on the price paid by long-distance bus the Spanish R&D Plan (project RTI2018-097434-B-I00). The usual users. disclaimer applies. 7 J. Asensio and A. Matas Research in Transportation Economics 100 (2023) 101310 Appendix Table A.1 Estimation results (clustered standard errors at contract level) Coefficient Std. Err. t-stat Constant 0.428 2.118 0.2 Ln(distance) − 0.081 0.061 − 1.33 Ln(income pc) − 0.156 0.157 − 0.99 Ln(population) 0.103** 0.021 4.82 Ln(speed) − 0.474** 0.126 − 3.76 Ln(bus/rail time) − 0.099** 0.021 − 4.75 Plane x ln(distance) − 0.012 0.008 − 1.59 Ln(attractiveness) − .023* 0.008 − 2.77 Tenders LOTT-mer − 0.131*** 0.014 − 9.26 LOTT-new − 0.307** 0.068 − 4.52 2007 − 0.197*** 0.031 − 6.39 2009 − 0.314*** 0.033 − 9.65 2011 − 0.214*** 0.033 − 6.38 2014 − 0.239*** 0.040 − 6.04 2016 − 0.769*** 0.067 − 11.41 Regional 0.056 0.049 1.15 R2 0.6258 Observations 221 Legend: *p < .05; **p < .01; ***p < .001. Table A.2 Estimation results excluding intra-regional services (clustered standard errors at contract level) Coefficient Std. Err. t-stat Constant 0.942 2.362 0.4 Ln(distance) − 0.0615 0.057 − 1.08 Ln(income pc) − 0.194 0.170 − 1.14 Ln(population) 0.101*** 0.022 4.51 Ln(speed) − 0.510** 0.167 − 3.06 Ln(bus/rail time) − 0.117*** 0.018 − 6.56 Plane x ln(distance) − 0.011 0.008 − 1.38 Ln(attractiveness) − 0.029*** 0.008 − 3.68 Tenders LOTT-mer − 0.144*** 0.009 − 16.38 LOTT-new − 0.267*** 0.050 − 5.34 2007 − 0.226*** 0.009 − 24.17 2009 − 0.332*** 0.029 − 11.33 2011 − 0.214*** 0.039 − 5.55 2014 − 0.261*** 0.033 − 7.91 2016 − 0.728*** 0.051 − 14.38 R2 0.6714 Observations 187 *p < .05; **p < .01; ***p < .001. References Beria, P., & Bertolin, A. (2019). Evolving long-distance passenger services. Market concentration, fares and specialisation patterns in Italy. Research in Transportation Economics, 74, 77–92. Aarhaug, J., Farstad, E., Fearnley, N., & Halse, A. H. (2018). Express coaches: An up-hill Beria, P., Nistri, D., & Laurino, A. (2018). Intercity coach liberalisation in Italy: Fares battle after liberalization? Research in Transportation Economics, 72, 82–91. determinants in an evolving market. Research in Transportation Economics, 69, Aarhaug, J., & Fearnley, N. (2016). Deregulation of the Norwegian long distance express 260–269. coach market. Transport Policy, 46, 1–6. Blayac, T., & Bougette, P. (2017). Should I go by bus? The liberalization of the long- AIReF. (2019). Spending review: Infraestructuras de transporte. Annex 1: La inversión y el distance bus industry in France. Transport Policy, 56, 50–62. stock de infraestructuras de transporte en España, autoridad Independiente de Cantillon, E., & Pesendorfer, M. (2006). Auctioning bus routes: The London experience. Responsabilidad Fiscal (AIReF), Madrid. Available at: https://www.airef.es/wp-con Combinatorial auctions, 22, 573–592. tent/uploads/2020/07/INFRAESTRUCTURAS/Anexo-1_La-inversion-y-el-stock- CNC. (2008). Informe sobre la competencia en el transporte interurbano de viajeros en autobús de-infraestructuras-de-transporte-en-Espa%C3%B1a.pdf. (Accessed 21 November en España. Madrid: Comisión Nacional de la Competencia. 2022). CNC. (2010). Informe de seguimiento del proceso de renovación de las concesiones estatales de Alexandersson, G., Hultén, S., Fearnley, N., & Longva, F. (2010). Impact of regulation on transporte interurbano de viajeros en autobús. Madrid: Comisión Nacional de la the performances of long-distance transport services: A comparison of the different Competencia. approaches in Sweden and Norway. Research in Transportation Economics, 29(1), CNMC. (2014). Informe sobre los pliegos tipo de las condiciones para los contratos de gestión 212–218. de servicios públicos de transporte regular de uso general de viajeros por carretera, INF/ Amaral, M., Saussier, S., & Yvrande-Billon, A. (2013). Expected number of bidders and DP/0004/14. Madrid: Comisión Nacional de los Mercados y la Competencia. winning bids: Evidence from the london bus tendering model. Journal of Transport De Haas, S. D. H., Herdold, D., & Schäfer, J. T. (2022). Entry Deterrence due to brand Economics and Policy, 47(1), 17–34. proliferation: Empirical evidence from the German interurban bus industry. Augustin, K., Gerike, R., Sanchez, M. J. M., & Ayala, C. (2014). Analysis of intercity bus International Journal of Industrial Organization, 83, 102844. markets on long distances in an established and a young market: The example of the Dementiev, A., & Han, H. J. (2020). A theory of deregulation in public transport. US and Germany. Research in Transportation Economics, 48, 245–254. Research in Transportation Economics, 83, Article 100953. Barrio, A. (2013). El proceso de renovación de las concesiones de transporte de viajeros (Vol. Dürr, N. S., & Hüschelrath, K. (2015). Competition in the German interurban bus 203). Transporte y Logística Terrestre. industry: A snapshot two years after liberalization. Competition and Regulation in Beck, A., & Walter, M. (2013). Factors affecting tender prices in local bus transport Network Industries, 16(3), 188–218. evidence from Germany. Journal of Transport Economics and Policy, 47(2), 265–278. 8 J. Asensio and A. Matas Research in Transportation Economics 100 (2023) 101310 Fageda, X., & Sansano, S. (2018). Factors influencing prices and frequencies in the Paulley, N., Balcombe, R., Mackett, R., Titheridge, H., Preston, J., Wardman, M., interurban bus market: Evidence from Europe. Transportation Research Part A: Policy Shires, J., & White, P. (2006). The demand for public transport: The effects of fares, and Practice, 111, 266–276. quality of service, income and car ownership. Transport Policy, 13(4), 295–306. Greco, S., Ishizaka, A., Tasiou, M., & Torrisi, G. (2019). On the methodological Preston, J., & Walters, J. (2020). Workshop 1 report: Models of mainstream public framework of composite indices: A review of the issues of weighting, aggregation, transport provision. Research in Transportation Economics, 83, Article 100924. and robustness. Social Indicators Research, 141, 61–94. Reynolds, T. J., & White, P. (2022). Forty years’ experience of deregulated express coach Groh, A. P., Von Liechtenstein, H., & Lieser, K. (2010). The European venture capital and services in Britain. Research in Transportation Economics, 94, Article 101181. private equity country attractiveness indices. Journal of Corporate Finance, 16(2), Ridderstedt, I., Pyddoke, R., & Nyström, J. (2019). Evaluating bids on price and quality – 205–224. implementations and effects in Swedish tenders of public transport. In Paper Hensher, D. A., Louviere, J. J., & Hansen, D. E. (2000). The use of mixtures of market and presented at Thredbo 16 - the international conference series on competition and experimental choice data in establishing guideline weights for evaluating ownership in land passenger transport, 25-30 August, Singapore. competitive bids in a transport organisation. Transport Policy, 7(4), 279–286. Thust, M., Kreling, C., & Fell, B. (2016). Modelling and observing the effects of long Holmgren, J. (2013). The efficiency of public transport operations–An evaluation using distance bus market liberalization in Germany. Transportation Research Procedia, 13, stochastic frontier analysis. Research in Transportation Economics, 39(1), 50–57. 81–89. Lieser, K., & Groh, A. (2011). The attractiveness of 66 countries for institutional real van de Velde, D. (2014). Market initiative regimes in public transport in Europe: Recent estate investments. Journal of Real Estate Portfolio Management, 17(3), 191–211. developments. Research in Transportation Economics, 48, 33–40. Merkert, R., Preston, J., Melkersson, M., & Link, H. (2018). Workshop 2 report: van de Velde, D., & Alexandersson, G. (2020). Workshop 2 report: Practical Competitive tendering and other forms of contracting-out: Institutional design and considerations in implementing different institutional regimes. Research in performance measurement. Research in Transportation Economics, 69, 86–96. Transportation Economics, 83, Article 100920. OECD. (2001). Competition Issues in road transport, policy roundtables, organisation for Vigren, A. (2018). How many want to drive the bus? Analyzing the number of bids for economic Co-operation and development (Vol. 10). Paris: DAFFE/CLP, 2001. public transport bus contracts. Transport Policy, 72, 138–147. OECD. (2008). Handbook on constructing composite indicators: Methodology and user guide. Paris: OECD Publishing. 9