Do Motorisation Statistics Reflect the Real Geography of Car Ownership in Poland?¹

For many years the annual European Commission reports (e.g. European Commission, 2018) have shown a relatively very high level of car ownership in Poland. According to the 2016 data, Poland was the sixth most motorised EU state (571 passenger cars per 1,000 inhabitants). It was by far the highest value among the post-socialist countries which joined the European Union in 2004. Poland’s car ownership rate amounted—in comparison to other socio-economic development indicators, and also in relation to other countries—to an above-average level, which necessitated specific recommendations in terms of transport policy. Likewise, international research on the drivers of change in car ownership clearly showed a non-standard course of this process in the country. In one of the most often quoted automotive articles (Dargay et al. 2007), where the authors compared car ownership changes and per capita national income in 45 countries in the years 1960–2002, the situation in Poland is shown to diverge considerably from the relationship between those two variables in the remaining states. It was a single case in the entire set (and the study concerned also a large group of developing countries). The analysis of car ownership changes and the income level in Poland in this period revealed that the income flexibility of car demand was more than twice as high as in the majority of European countries. This is certainly related to the development path Poland chose after 1989. However, what is equally important in shaping such a picture of the process (if not more important) is the quality of the Polish motorisation data used for research and international comparisons, which overestimate the car ownership rates.

For many years the studies on the car market have shown (e.g. Kublik 2005, 2013a, 2013b) that official data on the number of passenger cars do not reflect the actual state of this phenomenon. Professional specialists and car experts are quite well acquainted with this problem (it was also analysed by the Institute of Automotive Market Research SAMAR), but it has not been introduced to scientific discussion on motorisation so far, except in rare cases (e.g. Komornicki 2008, 2011; Menes 2018). This study attempts to fill the gap.

The article aims to identify the shortcomings of the statistics describing the size and structure of passenger car parc in Poland, and to formulate the implications of these limitations for geographical research. What made this objective possible was the creation of a new, increasingly reliable source of information, i.e. the Central Vehicle Register [PL: Centralna Ewidencja Pojazdów (CEP)] 2.0, which holds details about all vehicles in Poland. This type of data is used for automotive analyses, including their regional profiles (e.g. Lansley 2016). Such analyses, however, have not been performed in Poland yet.

The analysis presented in the later sections of the article was carried out on the basis of the data obtained from CEP on selected features of passenger cars (the year of the first registration in Poland, the year of production, permissible load capacity, the number of seats and the condition of a vehicle) and their owners (the district

District (powiat) – local territorial unit in Poland (between a commune and a province).

The article is composed of four sections. The first section shows the most significant shortcomings of motorisation statistics in Poland based on the review of related literature. Then, in the second section we describe the way of collecting data and keeping records of motorisation statistics in Poland. The in-depth analysis of the CEP data, which makes it possible to estimate the size and structure of selected limitations of motorisation statistics, is dealt with in the third section. Finally, we attempt to determine the implications of these shortcomings for geographical research.

To date, the motorisation data have been used in the investigations of a series of socio-economic phenomena and processes. The significant proportion of information on the number of passenger cars in Poland comes from the Statistics Poland [PL: Główny Urząd Statystyczny (GUS)] database. The data were accepted rather indiscriminately (with few exceptions), and while their weaknesses were discussed at conferences and professional meetings, they were less frequently analysed in scientific publications. The availability of the CEP data creates new analytical opportunities and allows estimating the size and structure of errors in the data used. When analysing the information on car ownership indicators, which are most often found in the Polish public statistics, one should focus on six major shortcomings.

The impact of the discussed shortcomings and their causes on the number of passenger cars registered in a given territorial unit (e.g. in a commune) is presented in Fig. 1.

Fig. 1

All the six shortcomings identified in the motorisation statistics significantly influence the research conducted in passenger car parc in Poland and the type of knowledge that is developing. Unfortunately, even CEP cannot fully recognise their entire scale and structure. This article will analyse the first three shortcomings in detail (i.e. ‘dead souls’, ‘cars with a grid’ and company cars). The remaining three shortcomings are far more difficult to be analysed in depth here, as they require different research techniques.

As was mentioned earlier, Polish car ownership research has been primarily based so far on the data published by GUS. This information on vehicles (including passenger cars) is made available as at 31 December each year. These data were obtained from various sources (Table 1). Until 2001 they had been kept by provincial agencies, called the Provincial Vehicle Register [PL: Wojewódzka Ewidencja Pojazdów (WEP)], and the data sent by car registering bodies were maintained by the following agencies over a certain periods: up to 1990 these were commune offices of a car owner’s domicile, then subregional administrative area (rejon) offices from 1990 to 1998, and from 1999 – districts’ authorities. The scope of the collected information about vehicles in the register included mainly basic technical data, i.e. a vehicle brand, a model, a type, the year of production, engine capacity, permissible load capacity, type of fuel, etc. In contrast, GUS, shared information about the number of cars (by type) at national and provincial levels.

The way of gathering and sharing car data was to change considerably after the establishment of the Central Vehicle Register. It was initially assumed that it would take place in the middle of 1999, pursuant to the Act of 1997 on the Traffic Code, introducing the provision for termination of WEP as of 30 June 1999. This fact caused widespread consternation because until that day the contractor for the new central base had not even been appointed (Centralna Ewidencja…, 2003). The Act of 1997 stated that from July 1, 1999, the Central Vehicle Register would start to operate (and thus the day before WEP would cease to function). However, it turned out quite impossible. Therefore, a few months later the provisions were amended (the Act of 31 March 2000) and WEP was reinstated alongside the need to update the regulations by heads of districts and cities’ mayors. Nevertheless, districts’ authorities were not obliged to enter the data on registered vehicles into WEP for almost a year (some of them publicly informed that they did not do it because there was no legal basis – cf. Bienias 1999; Burzyński, Bienias 1999). Given the circumstances, the continuity of the information collected was disturbed, and as a result, district and province databases varied in content.

In 2003, the Ministry of the Interior and Administration [PL: Ministerstwo Spraw Wewnętrznych i Administracji (MSWiA)] began consultations on the conception of the establishment, implementation and exploitation of a new IT system for the central registration of vehicles. The initial materials published by MSWiA revealed a series of problems which made the integration of district and province data difficult. This was so because local and regional records were kept in six different programming languages, seven operating systems and five database environments (some local units created systems based on their own, individual database solutions). There were also problems with the possibility of exporting local databases, the use of different dictionaries (codes defining particular features of vehicles), and no access to a LAN (Local area network) in about 1/5 of district authorities (Centralna Ewidencja…, 2003: 12). Notwithstanding, MSWiA signed an agreement with a contractor on the activation of the first CEP version until the end of December 2003. The database was activated on January 1, 2004; its assessments, however, published at the time in the trade and everyday press were very critical. One may even state that the first CEP version was very impractical because of the quality of data collected from local and regional databases. Furthermore, it coincided with the launch of the IT subsystem ‘Pojazd’ (the Vehicle), which was responsible for car registration in district offices as well as with a large number of new vehicles on the Polish market following Poland’s accession to the European Union. It turned out at the time that a relatively great deal of information about cars (concerning a set of about one million vehicles) had to be entered manually, because a newly created system was not able to implement the existing data (Samcik 2005).

The complete functionality of CEP was originally planned for the end of 2006. This date was postponed several times due to the need to build the related infrastructure (a bunker for storing copies of data from the system and a communication network allowing data to be read and updated in a real time). The integration of the CEP system operated by MSWiA with the ‘Pojazd’ system managed by the Ministry of Transport was also troublesome because of difficulties with the cross-sectoral cooperation. Eventually, it was only in 2009 that GUS started to publish CEP data (including those at district level), but this database was still impractical as was expected, which means that it did not integrate information from a number of different relevant data sets, including the database of the Insurance Guarantee Fund (PL: UFG, information about OC—liability car insurance) or vehicle inspection stations. This meant in turn that the verification of the number of registered cars would not be possible.

This state of affairs was to change finally after the launch of so-called CEP 2.0. In 2013, the Ministry of the Interior and Administration signed an agreement with the National IT Centre on implementing second-generation CEP, which was to be completed in 2016. This deadline was not met again and CEP 2.0 was launched as late as in November 2017. Since then, much more information on vehicles has been collected in this database, although its quality varies considerably. Today, the Central Vehicle Register is regulated by the amended Act on the Traffic Code (Journal of Laws 2020) and the Ordinance of the Ministry of Digitisation of 25 May 2018 on the catalogue of data gathered in the Central Vehicle Register (Ministry of Digital Affairs 2018). In the light of those two documents, the registry collects information on 60 different features describing technical data of each vehicle. Among them are the following: a vehicle brand, a model, a type, the year of production, engine capacity, permissible load capacity and the type of fuel. Some data, however, have poor information quality due to gaps and erroneous names. The weaknesses of CEP data limit their research use; however, it must be emphasised that despite those difficulties, this registry is a very detailed and precious source of data, making it possible to conduct many car ownership studies which are more reliable than before.

In the light of a rapid growth of statistical data, their increasingly easy acquirement and processing, one may observe a relatively low interest in their quality. This relates also to the application of over- or underestimated values of statistical indicators in geographical research, which may lead to a distorted spatial picture of the occurrence and dynamics of a given phenomenon or process, and as a result, to wrong interpretations and inappropriate practical measures. In this article, we have presented evidences that motorisation statistics do not fully reflect the real geography of car ownership in Poland.

The implications of the distortion of car ownership rates are twofold: direct and indirect. The direct consequences mean that the correction of the indicator’s value often drastically change the spatial distribution of the level (or dynamics) of car ownership and the position of a given territorial unit (country, province, district and commune) against other units. This discrepancy can be illustrated well by the comparison of Polish car ownership maps made using both GUS and CEP data, corrected by the authors of this article (Fig. 5). This correction consists in deducting end-of-life and company cars from the total number of registered cars and adding LGV-approved passenger cars (by criteria defined in section ‘Cars with a grid’ or LGV-approved passenger cars). For both variants of the car ownership rate we use a decile system, i.e. the set of districts was divided into 10 groups equal in number, ordered by the rate of cars per 1,000 inhabitants.

Fig. 5

The scale and spatial distribution of the discrepancy in the assessment of car ownership are shown in Fig. 6. It demonstrates the shifts of particular districts in the decile system after the authors’ correction of the car ownership rate, that is a change in the position of a given district against all such units in Poland. In the significant proportion of districts (256, i.e. about 66%), those changes have not occurred or are very small, and fall within the range of <−1; +1>. The most substantial shifts in minus can be observed in large cities (Sopot, Warsaw, Katowice, Wrocław and Poznań), which ‘are losing’ mainly because of deducting company vehicles, and in the group of districts with a large percentage of ‘dead souls’ and the relatively low number of ‘cars with a grid’ (e.g. the districts of Głubczyce, Przysucha, Sztum and Świdnica). An increase in the car ownership rate is noticeable among districts with reverse features, i.e. a small share of end-of-life vehicles and LGV-approved passenger cars.

Fig. 6

The indirect implications are linked to the fact that geographical research results concerning the level and dynamics of car ownership are used frequently in other broader studies related, e.g. to regional and local development factors, delimitation of peripheral areas, spatial accessibility as well as in planning and decision-making activities (e.g. construction projects of new transport infrastructure in areas with the greatest or lowest concentration of car ownership). The adoption of inadequate car ownership rates may lead to misguided transport investments. This is so because there are government studies (e.g. Strategia rozwoju…, 2013: 36, 53), which refer to misleading statistics of individual car ownership in Poland (but also to the overestimated number of LGVs and buses), and on that basis inaccurate diagnoses were provided. This has far-reaching implications. In recent years the opinion-forming media has produced texts that are completely divorced from reality, e.g. The most motorised countries in the EU. We rank high, but we drive junk (Forsal 2017), Poles are one of the most motorised nations in the European Union (Bankier 2018), Warsaw has almost three times more cars per 1,000 residents than New York does. Cars devour Polish cities (Pacholski 2018), which in turn have formed public opinion and the perception of transport problems in the country, regions, or cities.

The need to eliminate these shortcomings of motorisation statistics (especially when these shortcomings are relatively serious as in Poland) is also driven by the needs of an ever more rapidly developing market geography, including the geography of car market and car ownership. For instance, car dealers should be accurately informed in which regions the passenger car market is saturated and where there are particularly big ‘relative shortages’ of car supply. This knowledge in turn entails further decisions, concerning, e.g. location of car dealerships and other services related to their operation. The studies of geographers, making data about car ownership credible in spatial terms, their correct interpretation and recommendations for planning and decision-making, may contribute significantly to enhancing the prestige of our discipline and increasing its social importance.