Cost-benefit analysis is a formal analysis of the impacts of a measure or programme, designed to assess whether the advantages (benefits) of the measure or programme are greater than its disadvantages (costs). Cost-benefit analysis is one of a set of formal tools of efficiency assessment . Efficiency assessment refers to analyses made for the purpose of identifying how to use scarce resources to obtain the greatest possible benefits of them.
Cost-benefit analysis is a technique which is based on welfare economics. There are many textbooks explaining in detail the problems encountered in a cost-benefit analysis and how to solve these      . Only the main features of the technique are described here. The main steps of a cost-benefit analysis are as follows:
- Develop measures or programmes intended to help reduce a certain social problem (e.g. road accidents or environmental pollution).
- Develop alternative policy options for the use of each measure or programme.
- Describe a reference scenario (sometimes referred to as business-as-usual or the do-nothing alternative).
- Identify relevant impacts of each measure or programme. There will usually be several relevant impacts.
- Estimate the impacts of each measure or programme in “natural” units (physical terms) for each policy option.
- Obtain estimates of the costs of each measure or programme for each policy option.
- Convert estimated impacts to monetary terms, applying available valuations of these impacts.
- Compare benefits and costs for each policy option for each measure or programme. Identify options in which benefits are greater than costs.
- Conduct a sensitivity analysis or a formal assessment of the uncertainty of estimated benefits and costs.
- Recommend cost-effective policy options for implementation.
Brief comments will be given with respect to each of these stages.
Cost-benefit analysis is typically applied to help find efficient solutions to social problems that are not solved by the market mechanism. Typical characteristics of problems to which cost-benefit analysis is applied include 
- They involve public expenditures, often investments. Projects are sometimes financed by direct user payment, but more often by general taxation.
- There are multiple policy objectives, often partly conflicting and requiring tradeoffs to be made. It is assumed that policy makers want solutions that realise all policy objectives to the maximum extent possible.
- One or several of the policy objectives concern the provision of a non-marketed public good, like less crime, a cleaner environment or safer roads.
- It is assumed that an efficient use of public funds is desirable, since these funds are scarce and alternative uses of them numerous.
Road safety problems have these characteristics. Efficiency is a technical term in welfare economics. Without going into details, a measure or programme is a judged to be efficient if the benefits are greater than the costs. In principle, it will then be possible for those who gain from the measure to compensate those who lose from it, so that nobody becomes a net loser.
To identify relevant measures or programmes, a broad survey of potentially effective road safety measures should be conducted. A measure is regarded as potentially effective if it has been shown to improve road safety
– and has not already been fully implemented – or if there is reason to believe that it will improve road safety by favourably influencing risk factors that are known to contribute to accidents or injuries.
For each road safety measure, alternative options for its use should be considered. If the problem to be solved is bicyclist injuries, and the measure considered is bicycle helmets, alternative policy options could be:
- Do nothing; leave to each bicyclist to decide whether or not to wear a helmet.
- Conduct a campaign for bicycle helmets, while leaving their use voluntary.
- Make the use of bicycle helmets mandatory for children.
- Make the use of bicycle helmets mandatory for everybody.
These are distinct and mutually exclusive options. For very many road safety measures, however, options for their use are best conceived of as a continuous variable. Thus, one may convert 50 junctions to roundabouts, 51 junctions, 52 junctions, and so on. Most infrastructure-related road safety measures can be applied in very small gradual steps like this. These steps can be approximated as a continuous variable, since there would normally be thousands of junctions or thousands of kilometres of road that are candidates for the use of a certain road safety measure.
Policy options in cost-benefit analysis are always compared to a reference scenario and represent changes from that scenario. Often the reference scenario will be to do nothing, i.e. not introduce the road safety measure for which a cost-benefit analysis is performed. In some cases, however, one may foresee that a certain road safety measure will be introduced without any action from government. As an example, electronic stability control is now rapidly becoming standard equipment on new cars and will spread in the car fleet during the next 10-15 years. In such cases, the foreseen rate of introduction should be regarded as the reference scenario.
The most relevant impact of a road safety measure is of course changes in the number of accidents or injury severity. Some road safety measures may, however, have additional impacts on mobility (travel time) and the environment. If a measure has such impacts, they should be included in a cost-benefit analysis. One of the objectives of such analyses is to help make tradeoffs between different, and sometimes conflicting, policy objectives. Impacts that are relevant for all policy objectives must therefore be included.
All relevant impacts should first be estimated in “natural” units, for example number of accidents prevented, number of additional hours of travel, and so on. Then all impacts should be converted to monetary terms, applying monetary valuations of the various impacts. More will be said later about the economic valuation of road safety.
Cost-benefit analysis is designed to identify policy options for which benefits are greater than costs. According to the theory underlying cost-benefit analysis, a policy option should normally not be adopted if benefits are smaller than costs. It will, however, often be the case that costs and benefits are not known with certainty. An explicit consideration of uncertainty, as a minimum in the form of a sensitivity analysis should be part of any cost-benefit analysis. A case showing this is presented in section 8.