The opinions expressed in the studies are those of the consultant and do not necessarily represent the position of the Commission.
Promising road safety measures based on cost-benefit analyses
Table 4, taken from a recent report by Elvik, shows cost-effective road safety measures in Norway, according to an analysis of road safety policy. A total of 39 measures are listed in Table 4, covering a broad range. 6 of the 45 measures that were included in the cost-benefit analysis turned out to be cost-ineffective. These measures are not listed in Table 4. The table shows first order effects. A first order effect is the effect a road safety measure has when it alone is effective and it is not combined with another measure. First order effects cannot be added, but they can be compared.
It is seen that some of the most cost-effective measures, contributing to the largest reductions in the number of fatalities, are new motor vehicle safety features. It is not within the power of the Norwegian government to introduce these measures. The safety benefits resulting from these measures will therefore only be realised if the new vehicle safety features are offered as standard equipment by car manufacturers, or if agreement is reached within the EU or the UN Economic Commission for Europe to make these vehicle safety features mandatory. Some of the vehicle safety features, notably electronic stability control, seat belt reminders and cars attaining 4 or 5 star according to EuroNCAP are already penetrating the market quickly.
The analysis indicates that there is too little police enforcement. Fairly drastic increases of police enforcement are cost-effective. It is cost-effective to treble the amount of speed enforcement and to increase random breath testing (drink-driving enforcement) by a factor of five. Again, however, one may doubt if such drastic increases will actually take place, no matter how cost-effective they are found to be.
As noted in section 5, the current monetary valuation of road safety differs greatly between European countries. One might therefore expect the results of cost-benefit analyses to vary correspondingly. To see if this is the case, results of cost-benefit analyses made in a number of European projects in recent years have been compiled and compared. Measures that have been analysed in more than one project have been selected, as only these measures can provide information on the possibility of generalising the findings of cost-benefit analyses between countries. The following measures have been selected:
As far as traffic calming and speed reducing measures are concerned, Elvik (1999) found benefit-cost ratios varying from 9.7 to -0.4 for Great Britain, depending on the type of road. For all types of road considered together, the benefit-cost ratio was estimated to about 3.5.
Table 4 Cost-effective road safety measures in Norway
| ||Estimated reduction of the number of road users killed or seriously injured (first order effects)|
|Road safety measure||Benefit-cost ratio||Killed||Seriously injured|
|Road-related safety measures|
|Pedestrian bridge or tunnel||1.47||3.3||10.6|
|Converting T-junction to roundabout||1.86||1.9||6.1|
|Converting X-junction to roundabout||2.62||3.0||12.0|
|Roadside safety treatment||2.77||0.5||2.1|
|Reconstruction and rehabilitation of roads||1.57||1.0||3.2|
|Guardrails (along roadside)||2.53||1.3||5.3|
|Median guard rails on undivided roads||1.40||1.7||2.5|
|Median rumble strips (1 metre wide)||2.41||1.0||1.7|
|Horizontal curve treatments||2.37||1.4||3.4|
|Upgrading substandard road lighting||2.75||0.8||1.8|
|Follow up road safety inspections||2.48||3.1||5.3|
|Traffic signals in T-junctions||5.17||0.0||0.1|
|Traffic signals in X-junctions||3.95||0.2||0.8|
|Lowering speed limit on hazardous roads||14.29||3.2||4.7|
|Upgrading pedestrian crossings||2.36||5.4||12.7|
|Vehicle-related safety measures|
|E-Call (assuming mandatory from 1.1.2009)||1.61||4.9||0.0|
|Electronic stability control||3.98||34.5||81.2|
|Front and side air bags||1.01||14.9||29.2|
|Enhanced neck injury protection||20.25||2.3||23.0|
|Seat belt reminders||16.21||11.7||35.9|
|4 or 5 stars in EuroNCAP||1.24||13.7||49.1|
|Intelligent speed adaptation (ISA-systems)||1.95||43.5||126.0|
|Design of car front to protect pedestrians||4.52||1.8||19.4|
|Front impact attenuators on heavy vehicles||2.12||6.9||9.1|
Table 4: Cost-effective road safety measures in Norway, continued
| ||Estimated reduction of the number of road users killed or seriously injured (first order effects|
|Road safety measure||Benefit-cost ratio||Killed||Seriously injured|
|Enforcement-related safety measures|
|Section control (co-ordinated speed cameras)||1.58||0.9||2.2|
|Feedback signs for speed||2.35||1.4||2.5|
|Alcohol interlock for drivers convicted of drink-driving||8.75||7.5||19.6|
|Seat belt enforcement||2.44||5.7||17.5|
|Technical inspections of heavy vehicles||1.41||0.6||1.1|
|Service- and rest hour enforcement||1.45||1.1||1.9|
|Bicycle helmet law||1.02||1.3||2.4|
|Law requiring pedestrian reflective devices||3.49||5.6||11.8|
|Road user-related safety measures|
|Elderly driver retraining||1.85||0.2||1.0|
In Germany , the benefit-cost ratio of narrowing lanes and installing speed humps in residential areas was estimated to 17. Corresponding benefit-cost ratio were estimated to between 2 and 4 in Israel and around 1.1 to 1.2 in Greece . For Sweden  as well as for Norway , negative benefit-cost benefit ratios have been estimated for speed reducing measures in residential areas. Thus, the findings of cost-benefit analyses of this measure are somewhat inconsistent. Reasons for the inconsistency are not known, but one can speculate that residential streets in Norway and Sweden typically carry lower traffic volumes than in the other countries and have lower accident rates.
Daytime running lights have been found to be very cost-effective in all the analyses quoted above, except for one , with benefit-cost ratios typically ranging between 2 and 5. The assumptions leading to these results are questioned by Knight et.al .They argue that the assumption made in most analyses of a greater effect of daytime running lights on fatal and serious accidents than on slight injury accidents is weakly supported by available evidence from evaluation studies. Replacing it by an assumption of an effect of about 6 % reduction of daytime multi-party accidents at all levels of accident severity, Knight et al  find that benefits are smaller than costs. However, by slightly altering other assumptions made in the analyses, for example relying on the HEATCO recommendations for the monetary valuation of safety, benefits once more become greater than costs, even if a uniform effect of 6 % on daytime multi-party accidents is assumed. This example shows that sensitivity analyses should always be a part of cost-benefit analysis and that, in some cases, results are found to be quite sensitive to small changes in the assumptions made. On balance, it is more likely that the benefits of daytime running lights are greater than the costs than the opposite.
With respect to intelligent speed adaptation, all the analyses quoted above report that benefits are greater than costs. For this measure, therefore, there is perfect consistency in the findings of cost-benefit analyses.
Increasing speed enforcement has also been found to be very cost-effective in all analyses. It would seem that enforcement is an underutilised road safety measure in all of Europe. The same conclusion applies to random breath testing.
The cost-effectiveness of driver eyesight testing, on the other hand, has been found to vary substantially between the countries in which this measure has been analysed. More specifically, it appears to be rather ineffective in Norway and the Netherlands, but more cost-effective in Spain, the Czech republic and Switzerland. Reasons for these differences are not known.
The conclusion is that in some cases the results of cost-benefit analyses appear to be valid in many countries, while in other cases there are large differences. The lesson is that cost-benefit analyses should be performed in every country and that one should not uncritically assume that the results of a cost-benefit analysis made in one country apply to another country.