Collusive behaviour in the automobile sector in Spain. Quantifying consumer damages

Table 1. Correlation matrix.

	Price	Centimetres	Power	Fuel consumption	Speed	Leght	Width	Gasoline
Price	1
Centimetres	0.868	1
Power	0.911	0.919	1
Fuel consumption	0.726	0.793	0.815	1
Speed	0.755	0.760	0.866	0.676	1
Leght	0.493	0.539	0.525	0.425	0.560	1
Width	0.531	0.571	0.580	0.497	0.609	0.975	1
Gasoline	0.159	0.101	0.245	0.481	0.179	-0.174	-0.126	1
Source: own elaboration based on Cars magazine.

In addition, a dummy variable that considers high-end brands (High-end) has been included for BMW, Nissan, Mercedes, Volvo and Porsche. These brands sell more luxury vehicles and this consideration, from a marketing perspective, can be a decisive determinant in setting prices. An individualized regression analysis for each brand has not been ruled out as the number of observations was too small to obtain statistically significant results.

5. RESULTS

5.1. Descriptive analysis

In this section we refer to the structure of the automobile market in Spain. We will also introduce the evolution of production and prices throughout the collusive period to identify the possible anomalous behaviour of both variables. Subsequently, we offer the price estimates that would have prevailed without these collusive practices.

The analysis of the automobile market is defined through the dynamics of supply and demand. In a purely competitive environment, the price will be a consequence of the equilibrium between these forces. First, let’s address a description of the supply.

The automobile sector is one of the most dynamic industries in Spain. In fact, the automobile industry ranks second, just behind the agro-food manufacturing industry, with a total of 17 operational factories in 2013. In fact, Spain is second to Germany in production of automobiles in Europe. In 2015, the automotive sector represented 8.7 per cent of Spanish GDP, employing 9 per cent of the active population (table 2), according to the Spanish Association of Car and Truck Manufacturers (ANFAC). In addition, the sector generates considerable knock-on effects in other sectors of the economy, creating demand for intermediate metal products and components. A large portion of vehicle production is exported to hundreds of countries. In fact, the 86.9% of the total production is exported. Europe is the main destination for our exports, with Germany, Italy, France, and the United Kingdom accounting for 74.7% of our exported vehicles.

According to ANFAC, there were a total of 77 car brands in the Spanish market in 2013. Table 3 shows the total number of registered vehicles (exclusively passenger cars), as well as that corresponding to each of the sanctioned brands7. As can be observed, none of the car brands exceeded a market share greater than 11 percent in terms of car registrations. However, certain groups like the SEAT Group, for example, which includes SEAT and Volkswagen, capture a market share of almost 18 percent, positioning themselves as the market leaders.

In any case, the sum of the sanctioned companies would cover 81 per cent of the car market in Spain. The evolution of automobile production would therefore be practically explained by the production evolution of these brands.

In figure 2 a wider perspective including the data corresponding to the evolution of the manufactured passenger cars from 2003 to 2015 is adopted. If we consider the period in which the collusive agreement took place (2004-2013), we observe a decrease in the number of manufactured units, especially visible from 2008 to 2012. This reduction reflects, in part, the impact of the 2008 financial crisis, which affected not only the car market but the entire Spanish economy. The car sector was especially sensitive to the financial crisis as the lack of access to credit caused by the recession prevented families from financing the purchase of vehicles and companies their investment projects.

The Spanish automotive sector has an average production capacity that surpassed three million units in 2004. However, since then, the sector has experienced a degree of underutilization due to a decline in demand. In 2005, 90 percent of the capacity was utilized, but in the last five years, only 80 percent of the capacity has been reached. Vehicle production in Spain has undergone a cumulative reduction of approximately 32 percent (31 percent for passenger cars and 34 percent for commercial vehicles).

Figure 3 shows the growth rates of passenger car production and GDP. As can be seen, there are strong discrepancies between both variables. In the 2003-2007 period, when the Spanish economy was experiencing an economic boom, passenger car production underwent sharp variations (a very abrupt decline until 2005 and a recovery until 2007).

The strong impact of the 2008 financial crisis on the automobile sector caused the Government to intervene. Different measures were defined including direct public participation in some companies, the definition and negotiation of financial alternatives and promotion of vehicle purchases by consumers (Aláez-Aller et al. 2009). In any case, this high volatility, with little correspondence with the product evolution, could reflect the existence of some type of collusive behaviour in line with economic literature.

The problems faced by the Spanish automotive industry during this period cannot be attributed to intrinsic causes but rather to external factors, as our closest environment, Europe, has also had to deal with a similar situation. The European automotive industry has seen the closure of 87 factories and the loss of 500,000 jobs. The EU-27 experienced a cumulative decline of around 18 percent in vehicle production from 2007 to 2012.

Spanish vehicle production decreased by an average of 7 percent annually from 2007 to 2012, resulting in a loss of over half a million vehicles on average per year from 2008 to 2012. In the first half of 2013, production showed a slight recovery. While the United Kingdom and Germany experienced declines of 9 percent and 10 percent respectively in cumulative terms, each with an average annual reduction of 2 percent. The UK’s production reduction exceeded 300,000 vehicles between both five-year periods, while Germany’s reduction was less than 90,000 vehicles.

We have followed the same procedure with prices. In this sense, there are two main official statistical sources in Spain: the INE and those corresponding to the Ministry of Industry, Commerce and Tourism (Statistics on the manufacture of motor vehicles and bicycles). Given that the INE data does not separate passenger cars from vans, we have just considered the former. Figure 4 shows the evolution of the average price of passenger cars from 2003 to 2015. As we can see, the trend increases throughout the collusive period (2003-2013), except for the years 2008-2009, the start of the financial crisis. The increase in prices is, as mentioned, much higher than the increase in prices for the economy as a whole.

Despite the overall decrease in prices in the general economy as reported by the SNI the prices of passenger cars do not follow the same trend. Prices actually showed a significant growth of 4.98 percent between 2013 and 2015, as observed in table 4. However, this trend is not uniform across all price strata. The two lowest percentiles of the price distribution showed a decrease, while the two highest percentiles evidenced an increase, particularly pronounced in the last percentile. In conclusion, we can affirm that high-end luxury cars maintain a price growth, while lower-end cars experience a decrease in prices.

Figure 5 shows the evolution of the passenger car prices and GDP growth rates. Both variables follow a very similar trend until 2009. From this year, however, and especially until 2013, the behaviour of prices is quite anomalous, standing well above GDP growth. Once again, this evolution could be the result of collusive practice.

In relation to after-sales services, which determine the final price of the vehicle, brands have a distribution system for new and used automobiles and the provision of after-sales services based on a network of dealerships. These dealerships can be either official dealerships of the brand or independent dealerships. The dealership sells the vehicle to the final customer as an independent company on its own behalf. When it comes to new cars, the automobile brand recommends a selling price to the dealership and then they sets the final selling price freely. This practice of recommending a resale price is covered by the Category Exemption Regulation when the market share of each party to the agreement does not exceed 30 percent, as long as it does not involve a minimum or fixed selling price. However, the determination of the final selling price of the car is closely related to the remuneration policy established by the brand for its dealership network.

This is a capital-intensive business that has undergone significant reorganization and concentration due to the economic crisis of 2008, which resulted in a decrease in car demand. Between 2007 and 2012, the number of dealerships decreased by 20 percent, leading to a continuous loss of employment, dropping from 180,202 to 130,409. According to data from FACONAUTO. At the end of 2013, 359 independent dealerships disappeared, while brand-affiliated dealerships increased from 90 to 102.

As for after-sales services, which were also part of the collusive agreement through the After-Sales Forum, there is limited publicly available data regarding the economic activity of this sector at both the European and Spanish levels. In Spain, only the Annual Trade Survey by the INE provides aggregated data on motor vehicle maintenance and repair or the trade of vehicle parts and accessories. It is true that the prices of vehicle inspections and repairs carried out in the official network of each brand may have been affected by the information exchange. However, approximately 80 percent of the vehicles that undergo inspections or repairs are new cars, still under warranty, so the impact on consumers regarding a possible overpricing of these services does not exist.

Regarding the demand for automobiles, it comes from two different channels: private individuals or companies. The demand from companies includes fleets of private companies and those engaged in car rental or leasing. In terms of volume, private individuals represented 55.18 percent of sales in 2013, followed by corporate fleets accounting for 26 percent of sales, and car rental companies with 18.51 percent. Among all vehicles sold, there is a preference for diesel cars, which outnumber gasoline-powered cars.

The decrease in demand has been felt throughout the automotive sector during the years 2008 and 2012, coinciding with the economic crisis. It was further exacerbated by two successive VAT increases in 2010 and 2012, raising the tax rate from 16 percent to 21 percent, and the depletion of the budget for the “Plan 2000E” in May 2010. This decline resulted in a cumulative reduction of nearly 24 percent in the period 2007-2012, with an average annual decrease of over 5 percent in the EU-27 market, while the decrease in the eurozone was similar (-22.5 percent). The impact of the crisis on the European market exacerbates the problem for the Spanish industry, as our sector has a strong export orientation. During the five years of the crisis, exports decreased by almost 28 percent, although they grew by 6 percent in the first half of 2013.

A sharp drop in sales is observed for all vehicles, reaching its lowest point in 2012 (figure 6). Since then a recovery is observed in 2013, stimulated by Government approval in 2012 of the incentive plans for efficient vehicles (PIVE) to encourage the demand for cars with lower energy consumption. In spite of the positive results of the PIVE plan, resulting in the scrapping of many vehicles, the vehicle fleet has continued to age. The average age of our fleet in 2013 is over 11 years old.

Considering the type of vehicle and its characteristics, the demand can be segmented into cars in the low or mid-range, with lower prices and qualities, and high-end cars, also known as “premium,” targeting a population group with higher incomes. According to the ANFAC 2013 report, more than 75 percent of vehicles belong to the mid-range and low-range categories, while the remaining vehicles fall into the high-end and luxury segments

5.2. Econometric estimates

Next, the hypothetical price that would have prevailed in the absence of collusive practices is calculated. In the first place, the price of automobiles is estimated using 2015 data (year without surcharge). The estimation method used is ordinary least squares (OLS). This model provides us with unbiased estimates of the coefficients (as long as no relevant variables have been omitted and there is no causality or endogeneity between the dependent and independent variables). Furthermore, OLS predictive models provide unbiased estimates, even if the predicted coefficients are biased (Davidson & MacKinnon, 1993).

Once the independent variables coefficients have been estimated, they are used to obtain the counterfactual of the prices, that is, the prices which should have been set in 2013 (the year with the premium), following the model. Finally, the premium is calculated in percentage terms through a variation rate between the hypothetical price and the real price.

The hierarchical method was used to make the estimates using least squares regression (table 5). According to this procedure, a linear regression model is initially adjusted including just one explanatory variable, and subsequently the rest of explanatory variables were added8. If the R-squared in the second model is significantly higher than the R-squared in the previous model, this second one will be selected. In addition, the F statistic, when introducing a new variable to the model, is checked to observe if the new variables add explanatory value to the model. In our case, estimate (1) is the one that reports the best results. The introduction of the variable cubic centimetres and width do not contribute significance to the joint estimate and the Student’s t statistic of each variable were not significant. As shown, the explanatory power of the model is positive given that the R squared is 0.84, which implies a high goodness of fit.

As we can see, poser in the one of the determining elements of price. Additionally, the variable corresponding to high-end brands has a positive and significative effect on prices. In this sense, it seems that there is a prior segmentation of consumers who purchase luxury brands at a higher price. In contrast, the variable gasoline had a negative impact on prices at that time. The negative effect of maximum speed on price may not appear very intuitive, but can be explained by some relatively inexpensive cars with relatively high maximum speed.

With the coefficients of estimate 1, statistically the most appropriate option, the hypothetical price that would have been reached in 2013 without price collusion has been calculated. We use thus the coefficients obtained for 2015 and apply them to the reality of the explanatory variables in 2013 and we had the contrafactual price for each observation.

Table 6 shows the descriptions of the explanatory variables for 2013 and 2015. The explanatory variables do not present large differences in average, except for horsepower, which on average has decreased in 2015. The average price at which vehicles should have been sold (32,722.9 euros) is considerably lower than the price at which vehicles were sold (33,343 euros).

The average surcharge has been calculated for each observation, that is, for each car model. The premium is calculated as a percentage variation between the actual price in 2013 and the counterfactual price. The average of this percentage is the average surcharge, which is determined to be 10.39 percent. This represents the compensation that consumers who purchased a car during the collusive period should receive. This surcharge has been calculated using the estimation coefficients of Estimate (1), which is the most accurate. Additionally, surcharges for estimations 2 and 3 are also presented.

It is worth emphasizing, once more, as stated by the European Commission, that said surcharge is not a perfect reproduction of what would have happened without collusive practices, but rather an assessment. It is impossible to know with certainty and precision the reality that would have prevailed without collusive agreements. However, this problem is faced by all damage assessments that attempt to reproduce what would have happened if the unlawful conduct had not occurred.

6. DISCUSSION

We offer a consistent method to estimate the price consumers would have paid in the absence of collusive practices. The procedure chosen, Before and After, compares the price of the car during the collusive period (2013) with that corresponding to a later period (2015). Regarding this methodology, both the European Commission (2013) and the CNMC (2020) propose comparative and non-comparative methods to quantify damages and compensations. We have deemed it more appropriate to use the former, as they allow for the use of real market-specific data. Within the comparative methods, we have considered the “Before and After” method, given the difficulty of finding a market with similar characteristics to the Spanish one and, most importantly, a homogeneous statistical source. This method has been used in the literature in works such as those of Laitenberger and Smuda (2013), in the detergent market in Germany, Hastings (2004), in gas stations, and Bernheim (2002) in the vitamin C market, both in USA.

As for the percentage of overcharge, it is higher than those calculated in the works of Bernheim (2002), Hastings (2004) and Laitenberger and Smuda (2013). In any case, our results are in no way comparable since we are dealing with markets for different products in different regional contexts. It is feasible that considering a product that, in principle, has a higher income elasticity, a higher percentage of overcharge could be established. In fact, it has been observed that the percentage of overpriced is higher in those vehicles with higher prices.

Indeed, this percentage should be interpreted with caution because, as stated by the European Commission, it is not feasible to determine with exactitude what price would have prevailed in the absence of the collusive agreement. Nevertheless, our work allows for an approximation to be made, which is of great utility when determining compensations. To the best of our knowledge, there are no studies that have conducted a similar analysis to ours for this market in this period.

Additionally, while statistical regression analysis is often the preferred method for measuring cartel damages, it would be interesting to consider the use of non-comparative methods based on production costs and profit margins to estimate a scenario without infringement. This analysis could be relevant if there could be an overestimate or underestimate of the damages.

7. CONCLUSIONS

The existence of cartels, whether explicit or implicit, theoretically results in an increase in the prices of the products affected by the agreements and a reduction in the quantity demanded. While economic literature has demonstrated these effects for consumers, in practical terms, there is no consensus on how to estimate the loss of consumer surplus. This issue becomes evident when, in real life, there is a court judgment that proves the existence of an agreement between companies that negatively impacts consumers, and the courts have to decide the amount of compensation.

Estimating the loss of consumer surplus in cartel cases can be challenging due to the complexity of market dynamics and the lack of perfect information about how prices would have behaved in the absence of the cartel. Additionally, the behaviour of consumers and the elasticity of demand can vary, making it difficult to predict the exact impact on consumer welfare.

In legal cases involving cartels, determining the appropriate level of compensation for affected consumers can be a complex task for courts. The methodologies used to estimate consumer damages may vary, and various economic and statistical tools are employed to arrive at a reasonable estimate. However, due to the practical challenges and complexities involved, reaching a consensus on the precise calculation of consumer losses remains a challenging task in cartel litigation.

The difficulty in accurately determining the counterfactual scenario (the price that would have prevailed without the cartel) is a common challenge in cartel cases. Due to the lack of actual data on how the market would have behaved in the absence of the collusion, economists and experts often use various economic models and statistical techniques to estimate the potential competitive price levels and consumer damages.

Our paper estimates the loss suffered by consumers from the alleged restrictive practices of competition by 20 companies in the automotive sector between March 2004 and August 2013 (Resolution of the sentence, in file S/0482/13 of the CNMC). The exchange of information between companies would have impacted car sale prices fixing, as well as the homogenization of conditions and future commercial plans for the sale and post-sale of cars in Spain.

We offer a consistent method to estimate the price consumers would have paid in the absence of collusive practices. The procedure chosen, Before and After, compares the price of the car during the collusive period (2013) with that corresponding to a later period (2015). Our work also considers the evolution of automobile production and prices during the collusive period and detects certain anomalous behaviours. Specifically, variations in passenger car production differ considerably from the evolution of GDP and a much higher increase in passenger car price.

In the explanatory model, the cubic centimetres, the power of the car, fuel consumption, maximum speed, the fuel type, and the brand of the car, are identified as determining price variables.

We have obtained, using the hierarchical least squares regression technique, the coefficients corresponding to each determinant and apply them to the reality of cars in 2013. The cubic centimetres and width, as determinants of prices, were not statistically significant in our estimates. The key determinant of prices is the variable that captures high-end brands. Apparently, there is a prior segmentation of consumers who purchase luxury brands at a higher price. The explanatory power of the model, as a whole, is positive, as the R square is 0.84.

According to the model, the average price at which vehicles should have been sold is 10.39 percent lower than the actual sale price, which represents the amount that consumers who purchased a car during the collusive period should receive as compensation. Our results show, therefore, that the existence of collusive practices has led to a substantial increase in vehicle prices.

While our estimates may not be precise, they can still provide valuable insights for courts and regulatory authorities when deciding on appropriate compensation for affected consumers. These approximations are essential tools to address the damages caused by cartels and contribute to ensuring fair outcomes for consumers who have been negatively affected by anti-competitive practices.

Responsible reporting and conflict of interest

The authors declare that they have no conflict of interest.

This article does not contain any studies with human participants or animals performed by any of the authors.

Informed consent: None.

Both authors have participated in the preparation of the article: theoretical framework, descriptive and econometric analysis.

REFERENCES

Aláez-Aller, R., Bilbao-Ubillos, J., Camino-Beldarrain, V., & Longás-García, J.C. (2009). Reflexiones sobre la crisis de la industria española del automóvil y sus perspectivas. ICE: Revista de Economía, (850), 41-56.

Basso, L. J., & Ross, T. W. (2010). Measuring the true harm from price-fixing to both direct and indirect purchasers. The Journal of Industrial Economics, 58(4), pp. 895-927. https://doi.org/10.1111/j.1467-6451.2010.00437.x

Bernheim, D. (2002). Vitamins Antitrust Litigation. MDL, (1285), 99-197

Brander, J., & Ross, T.W. (2006). Estimating damages from price-fixing. Canadian Class

Action Review, 3, 335–369.

Boone, J., & Müller, W. (2012). The distribution of harm in price-fixing cases. International. Journal of Industrial Organization, (30), 265-276. https://doi.org/10.1016/j.ijindorg.2011.11.002

Comisión Nacional del Mercado de Valores (2015a). Resolución (Expte. S/0482/13 Fabricantes de automóviles). https://www.cnmc.es/sites/default/files/685749_68.pdf

Comisión Nacional de los Mercados y la Competencia (2015b). Guía práctica, Cuantificar el perjuicio en las demandas por daños y perjuicios por incumplimiento de los artículos 101 o 102 del tratado de funcionamiento de la Unión Europea. https://www.cnmc.es/sites/default/files/685749_68.pdf

Comisión Nacional de los Mercados y la Competencia (2020). Guía sobre la cuantificación de daños por infracciones de la competencia G-2020-03. https://www.cnmc.es/file/306138/download

Davidson, R., & MacKinnon, J.G (1993). Estimation and Inference in Econometrics, New York, Oxford University Press.

Davis, P., & Garcés, E (2010). Quantitative Techniques for competition and antitrust Analysis. Princenton University Press. https://doi.org/10.1515/9781400831869

European Commission (2013). Documento de trabajo de los servicios de la comisión. Guía práctica para cuantificar el perjuicio en las demandas por daños y perjuicios por incumplimiento de los artículos 101 o 102 del Tratado de funcionamiento de la Unión Europea. https://ec.europa.eu/competition/antitrust/actionsdamages/quantification_guide_es.pdf

Han, M. A., Schinkel, M. P., & Tuinstra, J., (2009). On the Overcharge as an Estimator for Total Chain Antitrust Damages. ACLE Working Paper Series 2008-08.

(Amsterdam Centre for Law and Economics, Amsterdam, The Netherlands). http://ssrn.com/abstract=1387096.

Harrington, J. (2002). Post-cartel pricing during litigation. Working Paper, (488). The Johns Hopkins University.

Hastings, J. (2004). Vertical relationships and competition in the retail gasoline markets: an empirical evidence from contract changes in Southern California. American Economic Review, (94), 317-28. https://doi.org/10.1257/000282804322970823

Laitenberger, U., & Smuda, F. (2013). Estimating Consumer Damages in Cartel Cases, ZEW, Centre for European Economic Research Discussion Paper, 13-069. https://doi.org/10.2139/ssrn.2333598

Oxera (2009). Quantifying antitrust damages, towards non-binding guidance for the Courts. European Commission.

Seixas, R., & Lucinda, C. (2019). Computing Cartel Overcharges: when theory meets practice. Estudos Econômicos (São Paulo). https://doi.org/10.1590/0101-41614935rsc

Van Dijk, T., & Verboven, F. (2007). Quantification of damages. In W.D. Collins (Ed.). Issues in Competition Law and Policy. IL: ABA Publications.

Wegner, A.C., Oberhammer, S., & Berger, A. (2016). Recent Competition Law Developments in the Automotive Industry. Journal of European Competition Law & Practice, 7(10), 696-710. https://doi.org/10.1093/jeclap/lpw070

Appendix 1

Period of cartel

Appendix 2

Brand car santioned

Brand car	Duration (months)	Turnover 2014 (€)	Santion (%)	Sanction (€)
Automóviles Citroen España S.A.	90	1,136,075,000	1.30%	14,768,975
B&M Automóviles eSPAÑ, S.A.	41	119,388,460	0.65%	776,024.99
BMW Ibérica, S.A.U.	59	1,147,397,086	0.70%	8,031,779.602
Chevrolet España, S.A.U.	90	13,858,000	1.00%	138,580
Chrysler España, S.L.	28	40,845	0.65%	265.4925
Fiat group automóviles España, S.A.	90	580,887,817	1.20%	6,970,653.804
Ford España, S.L.	90	6,744,944,000	0.30%	20,234,832
General Motors España, S.L.U.	90	380,462,847	0.80%	30,437,027.77
Honda Motor Europe Limited	52	203,108,411	0.30%	609,325.233
Hyundai Motor España, S.L.U.	41	464,749,000	0.95%	4,415,115.5
Kia Motors Iberia, S.L	69	414,844,000	0.50%	2,074,220
Mazda Automóviles España, S.A.	24	218,796,801	0.30%	656,390.403
Mercedes Benz, España, S.A.	12	2,379,329,985	0.10%	2,379,329.985
Nissan Iberia, S.A	62	287,061,039	1.10%	3,157,671.429
Peugeot España, S.A.	90	1,209,434,000	1.30%	15,722,642
Porsche Ibérica, S.A.	38	122,600,000	0.20%	245,200
Renault España Comercial, S.A.	90	1,516,951,831	1.20%	18,203,421.97
Seat, S.A.	84	1,334,205,000	0.95%	12,674,947.5
Toyota España, S.A.	90	824,477,429	1.05%	8,657,013.005
Volkswagen Audi España, S.A.	57	2,526,000,000	1.05%	26,523,000
Volvo Car España, S.A.	41	213,260,391	0.80%	1,706,083.128
TOTAL		25,262,037,566	0.71%	178,382,499
Source: CNMC (2015a).