Overview of the Motor Vehicle Inspection Market in Sweden
A periodic car roadworthiness test has been mandatory in Sweden since its introduction in 1965.Footnote 6 By law, vehicles are required to pass a mandatory periodic inspection to operate legally on the road.Footnote 7 The mandatory inspection includes both emissions and other safety inspections such as wheels and control system, drive system, brake system, communication and other. Vehicle inspection firms are not allowed by law to engage in any other business but inspection services, which means that their primary revenue comes only from the inspection business. Prices are not restricted by the regulator which allows inspection firms to set their prices based on market conditions.Footnote 8
A state controlled company, Bilprovning AB,Footnote 9 had a monopoly right to provide car inspection services until the government deregulated the market in July 2010. The reform opened the door for accredited private firms to provide inspection services alongside Bilprovning AB. The primary goals of the reform were to introduce competition to the market and, thereby, improve consumer welfare through increased geographical accessibility to the service, reduced prices, better service quality and longer opening hours.
The first private firm, Abesiktning AB, entered the market at the end of 2010. Another two private firms entered the market in 2011. Two more private firms entered the market in 2012.
To enhance competition further, the government decided to sell part of Bilprovning AB to a private firm. Accordingly, during 2012 a private firm, Opus Bilprovning AB, entered the market by buying 70 inspection stations from the government. During 2012 and after the sale of part of Bilprovning AB, the state and the other co-owners of Bilprovning AB agreed to split the remaining 144 stations between themselves.Footnote 10 As a result, the state continues to provide inspection services with 89 stations under the old company name, Bilprovning AB, and the other co-owners operated 55 stations under a new inspection company: Besikta Bilprovning AB.
All of the companies that operate in the market need to obtain accreditation from a government agency: the Swedish Board for Accreditation and Conformity Assessment (SWEDAC). The regulator of the market, Swedish Transport Agency, provides the rules and regulations that inspection companies should follow – such as which equipment and methods to use – and controls the competence of the inspection technicians. The Swedish Transport Agency has the responsibility to make sure that the regulations are not violated by inspection companies. The agency supervises the market by visiting the inspection stations and conducting statistical analysis on the information that is provided by the inspection firms.
Our article uses station-level monthly panel data over the period July 2010 to August 2015. The Swedish Transport Agency – the regulator of the market – provided us with the data that represent 22.5 million inspections of cars that weigh less than 3,500 kg. The data represent all inspections that were conducted between July 2010 and August 2015 by all licensed stations in Sweden.Footnote 11
The data include detailed information on the number of inspected vehicles and the percentage of vehicles that pass the inspection at the station level. The data also include information about each station’s date of entry and exact geographic address.
Using the data from the Swedish statistics bureau, we construct annual mean income at the municipality level. We also construct median car owner age and car age at the municipality level for each year. We also have data on the population size of each municipality.Footnote 12
Table 1 contains summary statistics for the main variables. Our dependent variable – Passrate – measures the fraction of total inspected cars that pass the inspection at a given station. Our main variable of interest – \(\#Stations\) – is the count of competitors that a station faces within its geographic market. The summary statistics table shows different versions of \(\#Stations\), which depends on the geographic market definitions. In the following section, we discuss the different approaches that we use to define geographic markets. We also discuss the benefits and concerns that are associated with each approach.
Measures of Competition
Estimating the effect of increased competition on pass rate requires an accurate measure of competition. In this paper, we measure competition for a given station by the number of competing providers within the station’s geographic market. An important element of this approach is to identify a station’s geographic market. In principle, the geographic market for a station should include all other competing providers to which the station reacts competitively.
In the main analyses, we use customer-level data to identify circular geographic markets. As a robustness test, we also identify geographic markets that are based on administrative boundaries (municipalities). The definition of circular geographic markets has two alternatives. Each of them is discussed below.
In the first approach, a fixed radius is chosen that represents the catchment area of customers for each station. Competition for a station is, then, measured by the number of competing providers within the area that has twice the radius of the catchment area. In this approach, all stations will have equal size circular geographic areas. For example, Bloom et al. (2015) define the catchment area for England’s hospital market with the use of 15-km fixed radius to all hospitals.Footnote 13 Bennett et al. (2013) use 0.2-mile radius around a facility to define a circular geographic market in the New York State vehicle emission test market.
In our paper, each station’s catchment area is defined by a 14 km radius.Footnote 14 Since stations with overlapping catchment areas can be considered as substitutes in the eyes of the car owners, we count each station’s number of competitors within 28 km. Figure 1 presents a graphical illustration of how the geographic market is defined based on a fixed-radius catchment area. Henceforth, we refer to this approach as fixed radius.
The fixed-radius definition, however, has its own limitation. It is reasonable to believe that the length of the radius that defines catchment area could be different across stations depending on local market characteristics. For example, one could expect that stations in urban areas would likely have smaller catchment areas than would stations in rural areas because of differences in population density.
The second approach tries to solve the limitations of the fixed-radius approach. This method uses the customers’ origin information to define station-specific catchment areas (Garnick et al., 1987). The Swedish Transport Agency provided us with data that contain detailed information about car owners’ registered addresses and the respective station that each owner chose to get an inspection service. For our purpose, we identified the latitudes and longitudes of the addresses of 458,405 car owners and the station that each owner chose.
We then calculated the road distance that each owner traveled for inspection service. By utilizing the distribution of these distances at the municipality level, we define the catchment area for each station.Footnote 15 The catchment area for each station is defined by the travel distance of the median customer to stations that were located in the focal station’s municipality.Footnote 16 Competition is, then, measured with the use of use the number of providers within the area that is double the radius of the stations’ respective catchment areas. Henceforth, we refer to this approach as variable radius.
Our main analyses are based on the fixed-radius approach. We, however, present robustness analyses that use both the variable-radius and administrative-boundary approaches. Administrative boundaries (municipal boundaries) and fixed-radius approaches offer a convenient way for defining the market because there is less demand for consumer-level data. Furthermore, political boundaries and the size of the catchment area are exogenous to factors that could influence pass rate. One limitation of the administrative boundaries approach is that a consumer’s choice of a station is not restricted by administrative region.
A major concern about using variable radius approach is that the size of the catchment area may not be exogenous to uncontrolled factors that could influence pass rate. A firm that promises higher pass rates (cet. par.) will attract more customers from farther away and thus have more competitors within its catchment area. For this reason and the other mentioned advantages, we choose the fixed radius approach as our preferred method in the main analyses.
Preliminary Data Analysis
The number of stations in Sweden increased over the sample period. Table 2 presents the evolution of the number of stations during the sample period. As of August 2015, there were eight companies with a total of 422 stations for light vehicles inspection. At the time of the deregulation, there were only the 190 state-owned stations.
Table 2 also presents the change in the number of competitors that the average station faces within its geographic market over time. At the end of 2010 and based on the fixed-radius approach (column 3), on average, each station competes with 2.5 other service providers within its geographic market. By August 2015 the average station competed with 8.7 other providers within its geographic market. The variable-radius approach in column 5 similarly indicates that stations faced increased competition over the years. In both approaches (columns 4 and 6), the number of competitors that the median station faced increased from 1 to 4 between 2010 and 2015, which is a threefold increase.
Table 3 presents the average and percentile breakdowns of the percentage of vehicles that passed inspections over time. While there was an increase in the number of inspection stations over the sample period, there was also an upward trend in the percentage of cars that pass inspections. The percentile breakdowns also show that the pass rates for the year 2015 (in the 50th, 75th, and 90th percentiles) exceed the pass rates of previous years for the corresponding percentiles.
Figure 2 presents early evidence on the relationship between pass rate and the number of competitors that a focal station faced over the period July 2010 to August 2015. In Fig. 2, the fixed-radius approach is used to define geographic market. The number of competitors that stations face is divided into five categories, from local monopolies to stations that face at least nine rival stations. The figure suggests that there is a positive relationship between the intensity of competition and the average pass rate. This relationship is also evident when the geographic market is defined using the variable-radius approach; see Figure 4 in the Appendix.
Pass Rate and Entry Pattern of New Entrants
One clear pattern of entry that we observe is that new stations enter municipalities where there are large populations. The stations that entered after the reform are located in municipalities where the mean (median) population size is 113,419 (60,422); whereas the incumbent (pre-2010) stations are located in municipalities where the mean (median) population size is 70,085 (27,297).
We also observe that those state-owned stations that later transferred to private owners are also located in municipalities where the average population size is larger than the population size of the municipalities where the stations that continue to be owned by the state are located. This may be an indication that the state-owned stations contribute more to ensuring accessibility in areas where private actors could be less interested.
To check on the pattern of the pass rates of the new entrants, we divided the ages of the new entrants (those that entered after the reform) into five categories: stations that had been conducting inspections during their first year, second year, third year, fourth year and finally during their fifth year after entry. The results show that the new entrants’pass rates do not show any noticeable trend over time, although it increases slightly (see Fig. 5 in the Appendix). We also compared the pass rates of the new entrants and incumbents. The unconditional mean comparison shows that the incumbents’ pass rate is 0.84 percentage points higher than the new entrants over the sample period (see Figure 6 in the Appendix).