Road safety is a major social aim. The countries that perform best in road safety base their most effective policies on an evidence-based, scientific approach. Countries may learn to improve road safety from their own experiences but also from systematic comparison with other countries. This study aims at providing an instrument that facilitates easy comparisons of the overall road safety situation between countries. Ideally, sets of indicators, describing the road safety outcomes and road safety policy performance are combined in one figure, a composite index, called in this report as the overall Road Safety Index (RSI) of a country. Thus performances on three levels of the target hierarchy for road safety (Figure 1) are systematically compared: 1) final outcomes (injuries and crashes), 2) intermediate outcomes (safety performance indicators such as drink driving, speeding, car safety) and 3) policy output (safety measures and programmes). Since social costs (top layer Figure 1) can be directly derived or calculated from the number of road users killed or injured, the present study did not seek information on social costs but is concerned with road safety outcomes and underlying intermediate outcomes, and policy output and input. In doing so, also the structural and cultural differences between countries should be taken into account as they form a different starting point for clustering the countries. In order to develop such instrument the current study focused on seven sub-objectives: 1. Select valid indicators for the final outcome, intermediate outcome, policy output and policy input layers of the pyramid. 2. Collect reliable data on these indicators. 3. Develop a method to combine the indicators of final outcome and intermediate outcome and policy output layers in one composite index. 4. Calculate the composite index for each layer from the available data. 5. Investigate the value of combining the composite layer-indices in one overall Road Safety Index. 6. Visualize the results for a set of European countries. 7. Develop a method to take into account structural and cultural differences between countries when comparing them on the preceding indices. Chapter 1 provides an overview of existing practices in comparing road safety results within the EU. A history of the background of the Road Safety Index is given as well as a short explanation of its nature, including thoughts on the acceptance of such an instrument by policy makers and politicians. The first SUNflower project, comparing road safety in Sweden, the United Kingdom and the Netherlands, aimed at a better insight into the development of policies and programs in the three countries to identify key factors to improve road safety. The three countries were chosen because, although they differ a lot, they have the best road safety level in the world. These road safety levels appeared to have been achieved through continuing planned improvements over recent decades. Their targeted policy areas had been similar, but their implemented policies differed at a detailed level. In a second study, called SUNflower+6, nine countries were studied using a similar method In the SUNflowerNext study, the concept of benchmarking was introduced to focus on learning from the best performing countries. The benchmarking concept, originated from quality control theories in the business sector, concentrates on improving performances by learning from others through identifying best performing countries, understanding why they are best performing, and by adapting outstanding practices from the countries which perform 'best-in-class'. Chapter 2 explains the method of the construction of the composite index. Composite indices are increasingly recognized as a useful tool in policy analysis and public communication. A large number of composite indices have been developed and applied in a wide range of fields during the last decades. In the road safety context, the development of road safety composite indices is also valuable in order to reduce the large amount of information and to provide a meaningful tool for national (or subnational) comparison and monitoring of road safety performance. Although the development of road safety composite indices is recommended, and some research efforts have already been devoted, care should be taken to ensure that the construction process of the index is transparent and follows sound conceptual principles. The construction of a Road Safety Composite Index involves several methodological stages. Having selected the set of indicators to combine in the road safety outcome index (Second below top layer, Figure 1), respectively the road safety performance index (Middle layer, Figure 1) and the road safety management index (One above bottom layer, Figure 1), and having obtained and prepared the indicator data, the next step is to apply the appropriate weighting schemes in order to deduce a weight for each indicator, and to compute an index score for each country subsequently. However, different weighting methods have their own advantages and limitations, and imply different end results. In general, no weighting system is above criticism. For this study, the technique of data envelopment analysis, known as the ‘benefit of the doubt’ approach, is chosen to construct the road safety composite indices, mainly due to the fact that the weights are retrieved from the observed data themselves, and more importantly, valuable information can be deduced, such as the identification of benchmark(s) for each underperforming country and the detection of aspects on which each country should focus. Chapter 3 describes the indicators that were chosen to represent the second top layer of the road safety target hierarchy (Figure 1), the Safety Outcome Indicators. The data for the safety pyramid outcome layer come mostly from the EC annual summaries, which are based on the Eurostat, CARE and other data sources. The summaries are annually produced which makes them a good platform for producing a Road Safety Index. The dataset was collected for 30 countries, most of the data concern 2008 except seat belt wearing rates which concern 2009 and car renewal rates which concern 2007. For the second top layer of the road safety pyramid, the following 7 indicators have been chosen: 1. the number of fatalities per million inhabitants, 2008; 2. the number of fatalities per million passenger cars, 2008; 3. the number of fatalities per 10 billion passenger-km travelled, 2008; 4. the share pedestrians among total fatalities, 2008; 5. the share pedal cyclists among total fatalities, 2008; 6. share motorcyclists among the total fatalities, 2008; 7. the annual average percentage reduction in fatalities, over 2001-2008. Chapter 4 describes the indicators for road safety performance, the intermediate outcomes in the road safety hierarchy (Figure 1). Safety performance indicators (SPIs) are measures (indicators), reflecting those operational conditions of the road traffic system, which influence the system’s safety performance. Basic features of SPIs are their ability to measure unsafe operational conditions of the road traffic system and their regular repeated measurement independent from the occurrence of specific safety interventions. SPIs are aimed at serving as assisting tools in assessing the current safety conditions of a road traffic system, monitoring the progress, measuring impacts of various safety interventions, making comparisons, and for other purposes. The chosen SPIs were: 1. the percentage of drivers above legal alcohol limit in roadside checks 2008; 2. the number of roadside police alcohol tests per 1,000 population 2008; 3. the daytime seat belt wearing rates on front seats of cars (aggregated for driver and front passenger) 2009; 4. the daytime wearing rates of seat belts on rear seats of cars 2009; 5. the average percentage occupant protection score for new cars sold 2008; 6. the average percentage score of pedestrian protection for new cars sold 2008; 7. the renewal rate of passenger cars 2007; 8. the median age of the passenger car fleet 2008. Chapter 5 deals with the road safety policy performance layer of the road safety pyramid. The Chapter reports on the actual insights into the effectiveness of road safety policy on the basis of available literature, and furthermore presents the results of a recently performed investigation by WP 1 on the validation of indicators of policy performance. The literature review showed that the institutional (road safety) management functions are almost without exception described qualitatively and need further operationalization. Moreover their impact will frequently depend on its quantity or intensity; this requires the assignment of quantitative values (categories). A preliminary investigation into potential road safety management indicators could not establish sufficient validity of the chosen indicators. It is concluded that little knowledge is readily available on valid indicators for policy performance, neither for institutional management functions nor for measures. Operational definitions are lacking as well as data on the topical occurrence of these conditions in the countries of Europe. Just one exception from literature can be mentioned, the effect of target setting on fatality reduction. Thus for the time being it will not be possible to construct a composite index for the quality of policy performance of a country and to value European countries in terms of this index. Chapter 6 pays attention to the Structure and culture layer of the road safety pyramid. The structural indicators consist of physical and social indicators that form the physical and functional structure of countries. The cultural level consists of the general norms, values and attitudes that may affect road safety, but that are not influenced by road safety policies. Both structural and cultural indicators can influence road safety but are themselves not influenced by road safety policies. The aim of the layer Structure and culture is to group countries in comparable classes. These indicators are used in later chapters to group countries into comparable classes. Comparable classes are constructed because it can be expected that countries learn more and more easily from similar countries than from countries which differ on physical and social characteristics. Also, countries might be more motivated to improve themselves if being the ‘best-in-class’ is considered to be within reach. Eight indicators were chosen to represent the structure and culture layer: 1. the share of people under 25 years old 2008; 2. the share of people over 65 years old 2008; 3. the population per 1 km2 of a country's territory 2008; 4. the percentage of population living in urban areas (>10.000 inhabitants) 2008; 5. the number of passenger cars per 1000 inhabitants 2008; 6. Share of goods vehicles in the vehicle fleet 2008; 7. the share of powered two-wheelers in the vehicle fleet 2008; 8. the GDP per head (based on EU27 = 100) 2008. In addition, it was also decided to explore the importance of Hofstede’s cultural dimensions in grouping countries. In Chapter 7 various forms of country grouping on the basis of structure and culture layer indicators were explored. Two groups of countries were identified using the four main country characteristics: GDP per head, motorization level, population density and the percentage of population living in urban areas. These two groups were stable across various classification methods. The key characteristics subdividing the countries into two groups were the indicators of motorization level and GDP per capita which are commonly known as characteristics of the level of a country's economic development. The first group includes 10 countries: Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Portugal, Romania, Slovakia, and, on average, is characterized by lower values of economic development. The second group includes the remaining 20 countries, that score generally higher, but also more diverse on the background country characteristics. Chapter 8 presents the results of an analysis which aimed at constructing a composite index based on road safety outcome indicators. By applying a data envelopment analysis, seven basic indicator values were combined into a composite index score for 30 countries. Two best-performing countries at the year under study were thereby identified, which were Iceland and Luxembourg. Furthermore, by obtaining a cross index score for each country, the countries were ranked and classified into five levels with respect to their road safety outcome. In total, nine countries were found to belong to the high level of road safety outcome, which are France, Germany, Iceland, Luxembourg, Netherlands, Sweden, Switzerland, United Kingdom, and Spain. Eight countries - Ireland, Portugal, Italy, Finland, Norway, Belgium, Malta, and Austria - were recognized as having a moderately high level of road safety outcome. In addition, six countries belonged to a medium level: Cyprus, Czech Republic, Denmark, Estonia, Latvia and Slovenia. Another six countries to a moderately low level: Bulgaria, Greece, Hungary,. Italy, Slovakia and Poland. Finally one country, Romania, belonged to a low level. Further re-estimating the composite index scores for the two separate country groups that were recognized earlier based on the background country characteristics in Chapter 7, most of the countries in the group characterized by lower values of the background characteristics belong to the last three levels of road safety outcome, i.e., medium, moderately low, or low. Regarding the remainder of countries, i.e., the 20 countries having higher values of the background characteristics, most of them belong to a high or moderately high level of road safety outcome,Considering the weight allocation provided by the DEA-based composite index model, for each country, the characteristics of relatively good and poor performance compared to other countries, can be identified (see Appendix I), thus providing a basis for planning road safety improvement efforts. It was found that a considerable difference in the countries' ranking appeared mostly depending on the inclusion or exclusion of the dynamic indicator (annual average percentage reduction in number of fatalities), whereas the addition of scope indicators (i.e., the shares of vulnerable road user fatalities) did not change the countries' ranking significantly. Taking into account the impact of the dynamic outcome indicator and the similarity of the results observed when including the additional scope indicators, it is recommended to further apply a composite index with respect to road safety outcome based on the whole set of seven indicators. Chapter 9 presents the results of an analysis which aimed at constructing a composite index based on road safety performance indicators (intermediate outcomes) of European countries. By applying a data envelopment analysis, eight basic safety performance indicator values were combined into a composite index score for 29 countries (Iceland was excluded from the analysis due to lacking SPI data). Moreover, by obtaining the cross index score for each country, the countries were ranked and further classified into five levels with respect to their road safety performance. Based on the safety performance index values, five countries were found to belong to the high level of road safety performance, which are Finland, Sweden, Norway, France and Ireland, in which Finland and Sweden are the two best-performing ones. Four countries — Germany, Netherlands, Spain, and Switzerland - were recognized as having a moderately high level of road safety performance. In addition, ten countries belonged to a medium level of road safety performance (Austria, Belgium, Cyprus, Czech Republic, Denmark, Estonia, Hungary, Luxembourg, Slovenia, United Kingdom), six countries to a moderately low level (Bulgaria, Greece, Lithuania, Malta, Portugal, Slovakia) and four countries (Italy, Latvia, Poland, and Romania) to a low level. Further re-estimating the composite index scores for the two separate country groups that were recognized earlier based on the background country characteristics in Chapter 7, the order of countries ranked in accordance with the safety performance index in the two separate groups repeated the order of countries received for the whole country set. However, once a more comparable country group was considered separately, a more realistic set of benchmark countries could be identified, especially for the country group with a lower level of the background characteristics, which in this study was Estonia and Hungary. Considering the weight allocation provided by the DEA-based composite index model, for each country, the issues of relatively good and poor performance, compared to other countries, can be recognized, providing policy makers with a basis for formulating road safety priorities for each country (see Appendix K). Finally, it is important to note that the selection of appropriate safety performance indicators requires periodic revisions. Apart from the SPIs developed in this study, other risk factors that have a strong relationship with road safety or a large contribution to road crashes and casualties, such as speed, road infrastructure, and trauma management, could also be incorporated in the future index research and corresponding indicators developed and data collected. In Chapter 10, statistical examinations demonstrate that the composite SPI index has a clear link with the composite final outcome index, but country rankings based on both indices are not identical. In this chapter first a general ranking of countries based on their two index scores was studied, and later this ranking was separately performed for two groups of countries that were distinguished in Chapter 7 on the basis of structure and culture indicators. It is possible to use a two-dimensional index for country comparisons, where countries are ranked simultaneously on the basis of the final outcome index and the SPI composite index. This method serves the main objective to compare a country with the ‘best of class’. The comparison makes clear which layer-index has to be improved in order to reach the performance level of the best performing countries. Further comparisons of the indicators composing the layer-indices reveal on which SPI(s) and/or on which final outcome(s) one should focus. In Figure 2 the country positions in accordance with their SPIs' and final outcomes' composite index values are presented. The dotted green lines indicate the boundaries of "moderately high" safety performance levels, according to the results of both analyses, thus, subdividing the area into four quadrants. The countries in the 2nd green quadrant (top right) are best performing on both indices, the countries in the other quadrants are less performing, on both indices (4th quadrant, bottom left) or only on the SPI index (1st quadrant, top left).The 3rd quadrant (bottom right) is empty. The figure enables any country outside the 2nd quadrant to compare itself with the best (moderately high) performing countries. A better final outcomes and/or SPI index value would enable it to move to the best quadrant. Further comparisons of the indicators composing the layer-index make clear on which SPI(s) and/or on which final outcome(s) one should improve. Chapter 11 looks back at the results achieved in every step of the total analysis strategy and gives conclusions on these with an emphasis on the practical use for policymakers. The chapter summarizes main results as follows. Indicators of final and intermediate road safety outcomes have been defined and currently available data have been collected for 30 European countries; the indicators on intermediate outcomes need further improvement. The technique of Data Envelopment Analysis (DEA) has been selected to construct a composite index. For each of the two layers of both final and intermediate outcomes a composite index has been constructed by using DEA. A visualisation was used to compare a country with the best of class according to the two composite indices in combination. The chapter recommends three areas for further research: 1. Improve the indicators of intermediate safety outcomes; this is necessary because a number of key valid indicators are currently missing (on speed, alcohol use, trauma management and roads). 2. Develop indicators and a composite index for effective and efficient policy performance. This research may aim at the valuation of either a country’s measures and programmes or of its institutional management functions, or both. 3. Construct an overall Road Safety Composite Index based on the three composite indices for the two layers of safety outcomes and the third layer of policy performance. This will require insight in the relative weights for the three layer indices. (Author/publisher)
Abstract