The other 94 road safety measures were not included in the impact study for various reasons. A recent road safety impact assessment for Norway is used as an example throughout the report (Elvik 2007A).
A normative model of road safety policy making
The impact assessment will generally be limited to those road safety measures that are considered realistic to implement. Therefore, an assessment of feasibility will always underlie a road safety impact assessment, albeit often implicitly.
Survey and selection of potentially effective road safety measures
Determining the framework for road safety impact assessment
Developing policy options – formal priority setting
Estimating expected effects of road safety measures
An estimate of the expected number of accidents for each “unit” of implementation of each road safety measure. An estimate of the expected effects of each road safety measure on the intended accidents or injuries.
The treatment of uncertainty
The effect of a road safety measure on traffic accidents is traditionally expressed in the expected percentage change in the number of accidents or the number of injured road users. In principle, the effects of road safety measures are likely to vary from case to case, depending on the many characteristics of the measures and the target groups to which they are applied.
Considerations relevant to policy choice and implementation
Example of a road safety impact assessment and a road safety programme
Screening of potentially effective road safety measures
Criteria for inclusion in a formal impact assessment
The effect of new measures cannot possibly be known in the same way as the effects of measures that have been used for a long time and have been thoroughly evaluated. The measure may of course be relevant in relation to other political objectives, but the impact assessment of traffic safety is based on impacts on safety.
Measures selected for a formal impact assessment
Patrol traffic (general enforcement) 802 No Ineffective measure Regulation of drinking and driving 803 No Fully implemented. Fixed penalties (traffic tickets) 809 No Ineffective measure Ordinary traffic tickets and imprisonment 810 No Overlaps other measures.
Forecasting baseline development
The effects of these changes on the number of fatalities and injuries in traffic accidents are assessed. A similar correction was made to determine the overall effects of the vehicle's safety features on the number of road users who were seriously or lightly injured.
Road safety targets in Norway
The other type of objective is an objective for reducing the number of dead or seriously injured road users. The government wants to reduce the number of dead or seriously injured road users by 50% in 2020 compared to the average annual numbers for the period 2003-2006. The aim is to reduce the number of road deaths from 250 to 125, and the number of seriously injured road users from 980 to 490.
General parameters for analysis
National transport plans, which include a long-term road safety programme, are drawn up every four years in Norway. The road safety impact analysis presented in this report is part of the preparation of the national transport plan. The risk premium only takes into account the relevant part of the risk, i.e. the risk that cannot be eliminated by having a diversified portfolio of assets.
Monetary valuation of relevant impacts
Regarding the discount rate, it has been estimated using the capital asset pricing method (Minken 2005). 1 police reported serious injury (adjusted for underreporting) police reported minor injury (adjusted for underreporting) 800,000 Travel time 1 vehicle hour passenger car travel 125 1 vehicle hour van travel 140 1 hour vehicle travel by freight truck vehicles 470 1 hour vehicle travel by bus (including passengers) 860 Vehicle operating costs Vehicle operating cost per kilometer – 1.30 Vehicle operating cost per kilometer – heavy vehicle 4.44 Environmental impacts Traffic noise, per vehicle km, large and medium cities 0.38 Traffic noise, per vehicle km, rural areas 0.00 Local air pollution, per vehicle kilometer, large cities 0.25 Local air pollution, per vehicle kilometer , small towns 0.11 Local air pollution, per vehicle kilometer, rural areas 00. Air pollution (carbon dioxide), per vehicle kilometer 0.12 Health impacts Uncertainty in crossing the road, for crossing 1.00 Uncertainty in walking or cycling in traffic mixed, per kilometer 2.10 Reduction of short-term sick leave, walking 1 kilometer 2.90 Reduction of short-term sick leave, cycling 1 kilometer 1.50 Reduction of serious illness, walking 1 kilometer 5.20 Reduction of serious illness, cycling 1 kilometer 2.60.
Service life of road safety measures
Constraints on road safety policy making
Efficient selection of sites for treatment
One of the objectives of the study was to assess the extent to which high-risk sites were selected for treatment. The average ratio between the observed crash rate at treated intersections and the normal crash rate at intersections in general was approximately 1.60. In short, the data describing the actual selection of treatment sites was strongly influenced by random variation and by unknown sources of site-to-site variation with respect to, for example, the costs of the measures and the degree of accident risk.
The logic of marginal analysis of road safety measures
The second stage of the marginal analysis of road safety measures is the determination of criteria for the selection of places for safety measures. The fourth stage of marginal analysis is to specify the relationship between traffic volume and the cost of road safety measures. It has been taken into account that the costs of road safety measures may vary depending on the volume of traffic.
First order effects of road safety measures included in analysis
The effects of the new front end and bumper standards on reducing pedestrian injuries were evaluated based on a study by Lawrence et al (1993). In most cases, these estimates were derived based on the Power model of the relationship between road speed and safety (Elvik, Christensen and Amundsen 2004). For some actions, the effects vary according to the "dose" of the action (this is the case with police enforcement).
Model for estimating expected affects on road safety
The results of this study were reanalyzed to obtain the score presented in Table 5.
Estimated first order effects of road safety measures included in analysis
Limitations of the current way of representing effects of road safety measures
An inventory of sources of uncertainty in road safety impact assessments
The possibility of estimating the contributions of various sources of uncertainty . 41
A solid curve has been drawn by hand in the figure to show the main thrust of the findings. The surrounding dotted curves summarize all data points except for a single peripheral data point that is identified in the Figure. When enforcement decreases (values to the left of 1 on the abscissa), there is always an increase in the number of accidents.
An exploratory analysis of models for estimating the combined effects of road
Here, the study is of interest mainly because it enables a comparison of the effects of 1, 2 or 3 road marking treatments. The study used a before-and-after control for regression to medium and long-term trends. For the study by Brüde and Larsson (1985), the ordinary residual model best fits three or four treatments.
Uncertainty of monetary valuation of impacts of road safety measures
When simple (unweighted) average estimates of the effect are used, the common residuals model best predicts the effects of three or four treatments, while the dominant common residuals model best predicts the effects of two or five or more treatments. If mean estimates of the effect are weighted inversely by the sample variance of the individual estimates, the dominant residuals model predicts best for all cases of multiple treatments. Although no model is clearly superior to the other, the dominant common residuals model appears to be somewhat preferred.
Uncertain duration of effects of road safety measures
In a comprehensive meta-analysis, de Blaeij et. 2003) summarized 95 estimates of the value of statistical life in traffic safety. A meta-analysis was performed to identify the sources of variation in the estimated statistical life expectancy. Given the current state of knowledge, it is difficult to fully quantify the uncertainty of monetary valuations of non-market goods in cost-benefit analyzes of road safety measures.
The estimation of compound uncertainty: an example
However, since the main purpose of this exposition is to demonstrate the logic of the method, this will not be done. The expected number of injuries prevented is multiplied by the social costs of road injuries and by a constant that reflects the time horizon of the analysis and the discount rate. Regarding the costs of traffic accidents, it is concluded that the standard error of the mean for the average costs of a police-reported injury accident is equal to 16.2% of the average costs.
Considerations included in, and excluded from, formal analyses of efficiency
If a particular public agency has a large range of cost-effective road safety measures under its jurisdiction, more than its current budget allows, the result of the efficiency analysis is that its budget is too small and should be increased. Public acceptability: An effective road safety measure must be popular or have public support. The following section discusses some aspects relevant to the implementation of road safety programs developed through performance analyses.
Issues regarding implementation of road safety programmes
- Empowerment
- Reallocation mechanisms
- Competing incentives
- Social dilemmas
- Public acceptance
However, national governments may not have the power to implement all cost-effective road safety measures. Road safety measures on the local road network may therefore be beyond the control of national governments. It is therefore relevant to collect information about public acceptance of various road safety measures.
Road safety impact assessment for Norway 2010-2019
It is close to a constrained optimization, but actually departs from optimality by increasing the use of some road safety measures to the point where the total benefits equal the total costs. At this point, the marginal benefits will be less than the marginal costs, but it can be argued that society as a whole is better off, since in principle the excess benefits in inframarginal projects (i.e. projects for which the marginal benefits clearly exceed the costs marginal) can be used to "cross-subsidize" the deficit in ultramarginal projects (ie projects for which the marginal benefits are clearly less than the marginal costs). Therefore, although this policy option departs from strict optimization, it is, in a sense, still faithful to the principles of welfare economics by respecting the criterion that the winners must be able to compensate the loser and still maintain a net benefit.
Main results of alternative policy options
- Optimal use of road safety measures
- Constrained optimal use of road safety measures
- Continuation of present policies
- Strengthening present policy
The number of seriously injured road users can be reduced from 1109 without the program to 652 if the program is fully implemented. The number of slightly injured road users can be reduced from 12,650 without the program to 9,942 if the program is fully implemented. The number of seriously injured road users in 2020 is estimated at 691 and the number of slightly injured road users at 10,551.
Comparison of policy options
Policy option D is more ambitious and involves a drastic increase in spending on some road safety measures. This option leads to a decrease in the number of road users killed or injured, which is almost as great as for option A, optimal use of road safety measures. Policy option A, optimal use of road safety measures, provides the greatest overall benefits but does not have the most favorable benefit-cost ratio.
Implications for private and public expenditures
- Road-related measures
- Vehicle-related measures
- Enforcement-related measures
- Road user related measures
- Total costs
In policy option C, costs are estimated at NOK 153 million for investments and NOK 59 million for operating costs. Expenditures for road-related measures are assumed to increase during the period 2010-2019 by equal amounts each year (10 years). Expenditures for vehicle-related measures are assumed to have accumulated in identical amounts over the years).
Assessing uncertainty
In all policy options, the reference value for the number of deaths – the number expected to occur in 2020 if no new road safety measures are introduced – is 285. Number of deaths affected by policy option (A): 285 Variance of number of deaths affected by policy option ( Var(A)): 335. The standard error of the number of deaths prevented is the square root of the variance, which equals 31.0.
Road safety impact assessment: useful or window dressing?
- Competing criteria for priority setting
- Efficient selection of sites for treatment
- Competing incentives
- Social dilemmas
- Public acceptance of road safety measures
- Power and path dependence
The (simple) average benefit-cost ratio for road safety measures where more than 40% of the mortality reduction benefits pedestrians or cyclists is 2.28. A previous analysis of the actual selection of locations for road safety treatment in Norway (Elvik 2004A) suggests that the current selection is not maximally efficient. Are any of the cost-effective road safety measures identified in this impact assessment likely to create social dilemmas?
Prospects for improving road safety in Norway: a summary
Based on the 45 road safety measures included in the impact assessment, four policy options have been developed for road safety policy in Norway. Estimates of the likely impacts of road safety measures show that the 2020 policy objectives are unlikely to be achieved. How would setting policy priorities based on cost-benefit analyzes influence the provision of road safety?
