To summarise, the current state of the art with respect to the use of cost-benefit analysis to set priorities for road safety measures can be characterised as follows:
• Cost-benefit analysis comprises many of the policy objectives that are relevant for road safety policy, but is unlikely to comprise all relevant policy
objectives. In particular the objectives of balancing regional development and reducing the differences in accident rate between protected and unprotected road users are unlikely to be given due weight within the current framework for cost-benefit analysis.
• It is not clear to what extent the objective of reducing road user insecurity, as well as insecurity among residents along a road is part of the accident costs and thus included in current cost-benefit analyses.
• According to the cost-benefit analyses made in this report, measures that have a potential of reducing the number of fatalities by nearly 25% are not efficient, in the sense that benefits are smaller than costs. Abstaining from introducing these measures by referring to cost-benefit analyses is inconsistent with the ethical principles underlying Vision Zero.
11.4 Constraints on Road Safety Policy Making – the Power of
infrastructure, public transportation level-of-service and fares, overall travel demand, modal choice, fuel and vehicle tax rates, size and structure of vehicle pool, driver’s license penetration rates, etc). Most importantly, many of these factors are strongly associated with aggregate exposure, i e with the total volume of activities exposing the members of society to road accident risk.
Fourth, the accident statistics depend, of course, on the system of data collection.
Accident underreporting is the rule rather than the exception. Changes in the reporting routines are liable to produce fictitious changes in the accident counts.
Fifth, accidents counts, much like the throws of a die, are strongly influenced by sheer randomness, producing literally unexplainable variation. This source of variation is particularly prominent in small accident counts. For larger accident counts, the law of large numbers prevails, producing an astonishing degree of long-run stability, again in striking analogy with the dice game.
Finally, accident counts are susceptible to influence – and, indeed, influenced – by accident countermeasures, i e measures intended to reduce the risk of being involved or injured in a road accident, as reckoned per unit of exposure.
Although generally at the centre of attention among policy-makers and
practitioners in the field of accident prevention, this last source of influence is far from being the only one, and may not even be the most important. To effectively combat road casualties at the societal level, it appears necessary to broaden the perspective on accident prevention, so as to – at the very least – incorporate exposure as an important intermediate variable for policy analysis and intervention.”
In the analyses presented in this report, exposure has generally been taken as an exogenous factor, with the exception of increasing taxes on the use of motor vehicles in order to match the level of external costs caused by the use of motor vehicles. Figure 18 is an attempt to sort out the relative contributions of some major factors affecting the number of road accident fatalities in Sweden, inspired by the taxonomy proposed by Fridstrøm.
The first factor which is outside the control of any agent, is random variation in the count of fatalities. The contribution of this factor has been assessed as 1.96 times the square root of the mean count of fatalities during 1994-1998. This comes to 46, which is 8.3% of the count of 554 fatalities. This represents the upper limit of the explanatory power of any model designed to explain the variations in fatality counts (in time or space). Any multivariate model with a multiple squared correlation coefficient exceeding the value of 0.917 (1 – 0.083) would be over fitted and would include a spurious explanation of random
variation as well as systematic variation.
In chapter 2, the population attributable risks of 20 risk factors was estimated.
Although incompleteness in the data set precluded the use of a multivariate technique, the combined estimated attributable risk came to 0.885. This means that, in theory, if all the risk factors could be eliminated, the number of fatalities could be reduced by 88.5%.
8,3 % 2,7 %
11,6 %
24,5 % 38,3 %
14,6 %
Random variation in count of fatalities (95% limits)
Risk factors not identified as a road safety problem
Risk factors not influenced by any known safety measure
Measures whose benefits are smaller than costs
Priority given to other policy objectives Expected effect of current road safety policy by 2012
Figure 18: A taxonomy of factors affecting the power of government to reduce traffic fatalities. Percentage contribution of each factor
The difference between the proportion of systematic variation in fatality counts, and the combined attributable risk of the risk factors included in the analysis (0.917 – 0.89 = 0.027) can be interpreted as an estimate of the proportion of fatalities attributable to risk factors that have not been identified as road safety problems. This may include a large number of risk factors, each of which has a very small effect on the number of fatalities.
The remaining proportion of fatalities, 0.89, is attributable to risk factors that may be partly subject to control by means of safety measures. It has been estimated that the maximum potential for reducing the number of fatalities is 77.4% (or 0.774 as a proportion). This means that a complete removal of all risk factors is impossible. In fact, 11.6% of fatalities (0.89 – 0.774) can be attributed to risk factors not amenable to control by means of any of the road safety measures considered in this report.
If one were to rely only on measures whose benefits are greater than the costs, the number of fatalities could be reduced by 293, or 52.9%. By refraining from using more expensive measures, that is measures whose benefits are smaller than the costs, the prevention of 24,5% (0.774 – 0.529 = 0.245) of fatalities is given up as being too expensive. This can be interpreted as the limit imposed on accident prevention by relying on cost-benefit analysis.
Many people may find it unacceptable to abstain from preventing nearly 25% of fatalities on account of a rather abstract economic line of reasoning, if it is in principle possible to prevent these fatalities. No claim is made in this report that relying on cost-benefit analysis is the only sensible approach to policy making.
Indeed, a number of serious objections can be raised to cost-benefit analysis.
However, the biggest problem in current road safety policy is not that an
excessive reliance on cost-benefit analysis is used as an argument for not applying effective road safety measures. A far bigger problem is that current priorities are ineffective, which means that a greater reduction in fatalities could be attained if priorities were set according to the results of cost-benefit analyses.
Figure 18 shows that the expected effect of current road safety policy is a
adverse effect of traffic growth). By using all measures whose benefits are greater than the costs, this could be increased to 52,9%. Hence, the priority currently given to other policy objectives – other than those that are formally included in cost-benefit analyses – implies that the Swedish government is refraining from using measures that would reduce the number of fatalities by close to 40%, at no additional net cost to society. By definition, all these measures provide benefits that are greater than costs.
It is surprising indeed that road safety policy is so inefficient in a country that has officially adopted a long term vision of zero fatalities and zero serious injuries as the only morally defensible ideal that can be set for road safety. By launching Vision Zero on the moral high ground, justifying it by means of set of ethical principles claimed to be self evident, the Swedish government has implicitly branded those who take exception to this point of as immoral cynics, who are prepared to sacrifice human life for the sake of an abstract economic principle.
But the Swedish government itself fails to live up to the moral standards it has set for road safety policy. There seems to be a growing distance between the lofty ideals of Vision Zero and the realities of a road safety policy that is not delivering any reduction at all in the number of road accident fatalities.