Entwicklung eines Verfahrens zur Generierung eines Safety Performance Indikators aus der Bewertung von Euro NCAP

The European New Car Assessment Programme (Euro NCAP) is a consumer protection orientated programme for the safety assessment of – as a rule – new car models. The programme was established in 1997, since 2009 it has consisted of 4 Boxes (1. Adult Occupant Protection; 2. Child Occupant Protection; 3. Vulnerable Road User Protection; 4. Safety Assist Systems).
The main objective of this project was to transfer the test results from Euro NCAP onto the overall car fleet wherever possible and to use this to form a Safety Performance Indicator (SPI) for the vehicle stock over several consecutive years (time series). A further objective was to investigate whether or not a relationship exists between the Euro NCAP assessment of vehicles and accident occurrences. For this purpose, in addition to literature analyses, statistical models were estimated on the effect of vehicle safety expressed by the SPI on the corresponding number of persons injured in accidents taken from the official road traffic accident statistics.
A fundamental step in the development of an SPI of vehicle safety consisted of making the Euro NCAP test results from the various years comparable as far as possible with regard to the test procedures which had changed in the course of time. For this purpose, a project group was formed consisting of experts from the Federal Highway Research Institute (BASt) in the fields of active and passive vehicle safety which had the task of quantifying, for each Box, the extent to which these changes to the test procedures affect the vehicle assessment. The test conditions of the year 2020 were used as reference.
The second fundamental step in generating a Safety Performance Indicator consisted of linking (matching) the – newly calculated – Euro NCAP result data for the individually tested makes and models to the Central Vehicle Register inventories (1.1.2014 to 1.1.2020).
The matching of the Central Vehicle Register inventory data with the Euro NCAP assessment data was carried out using a complex algorithm which is based, in essence, on the characteristics of make code, model code and year of initial registration.
As a result, on average over the seven years observed here (reference dates: 1.1.2014 to 1.1.2020), around 70% of the newer vehicles (initial registration year from 2009) were able to be allocated to a Euro NCAP assessment. The remaining cars in the Central Vehicle Register inventory were assigned the missing assessment using an imputation procedure.
A total of four (Box-specific) Safety Performance Indicators were formed which are based on the newly calculated and standardised vehicle assessments from Euro NCAP. The specified indicators are mean values of the corresponding safety assessment of the cars recorded in the Central Vehicle Register. Using these four indicators, an Overall Safety Performance Indicator was then also calculated using weighted average. Thus, as a result, time series for the four Box-specific SPI values as well as the overall SPI value are available.
The core result of the analyses of the SPI time series is that all indicators show an increasing time trend. This is a clear indication that the safety standard has not only continuously risen in recent years for new vehicles, but also in terms of the overall car fleet.
The highest indicator values can be found in the SPI for Box 2 (Child Occupant Protection), they turn out lowest with regard to the Safety Assist Systems (SPI for Box 4). If the indicators are additionally grouped into car segment, then SUVs, followed by ATVs, commercial vans and the luxury class show the highest values in the overall SPI value. The fact that SUVs have the highest SPI is certainly also related to the fact that this is a relatively new segment in which the share of older vehicles is comparably low.
As part of the accident analysis, log-linear regression models were estimated in order to determine the effect of the four SPIs on each of the corresponding accident characteristics (injured car occupants, pedestrians and cyclists, cars involved in accidents). In addition, the relationship between the overall SPI value and the resulting monetary economic accident costs was analysed. The statistical assessments on the association between SPI and accident occurrences show, in all cases, that a higher value of the corresponding Safety Performance Indicator is accompanied by a lower number of persons injured in accidents or cars involved in accidents. With regard to accident costs, the results here were a (significant) reduction of the accident costs by around 0.7%, when the overall SPI increases by 1%.
In addition, in the concept of the project, the fundamental prerequisites for an uninterrupted continuation of the SPI time series in the following years were already created.