A statistical model to compare road mortality in OECD countries
Introduction
Life and death depend on destiny. (Confucius)…… But, as far as road safety is concerned, they do not. Our planet shelters about 6 billion people, more than 22 million km of roads, 470 million passenger cars and 145 million station wagons, vans and trucks2. One third of motorised vehicles move in the USA and another third in the European Union (EU) (ONISR, 1996, International Road Federation, 1995). According to the World Health Organization, 600 000 people die and 15 million are injured every year in traffic accidents3. Two thirds of the casualties occurred in developing countries. Seventy-five percent of casualties in these countries are vulnerable road users such as pedestrians or two-wheelers (Downing, 1995).
The EU counts about 368 million inhabitants and 183 million motorised vehicles. A total of 1 200 000 injury road accidents and 45 700 fatalities occur every year, slightly more than in the USA (42 300 in 1995) with 260 million inhabitants and 198 million motorised vehicles.
In most countries, accident economic losses reach 1 or 2% of GNP (Ross and Ray, 1995). In 1997, the European Transport Safety Council estimated the total cost of transport accidents in Europe at 166 billion Euros (ETSC, 1997). Ninety-seven percent of these costs, i.e. 162 billion Euros, were directly related to road transport.
Road safety has been increasing in industrialised countries for 25 years and this increase shows that political willingness and counter-measures produce results (Broughton, 1991, Downing, 1995). Improvements in road safety and the numerous results in the evaluation of safety measures led some countries to compare their performance and to use these comparisons as a tool to evaluate the effectiveness of their safety policy. It is then rightful to wonder whether the common indicators (fatalities and fatalities rates) are adapted to evaluation.
Numerous factors influence a country's safety level. These factors are concerned with road safety policy, distribution and crashworthiness of the fleet, road network characteristics, human behaviour and attitudes, and so on (Brenac, 1989, Brühning, 1995, Sivak, 1996). Among these factors, some are exogenous to road safety pure performance (safety actions, traffic safety policy and social acceptance), and some are endogenous (Poppe, 1995).
The evaluation of safety actions by international comparison consists in evaluating the performance of a country, i.e. the effectiveness of endogenous factors in each country, the exogenous factors being neutralised. The performance must be understood as the ability of a road safety policy to be effective and the ability of a population to accept and respect this policy. This performance measures the pure success of a country in its struggle against road mortality, independently of exogenous factors that influence the road safety level but which can not be modified by safety policy.
For example, the age pyramid is exogenous to safety performance: in all countries, the accidental risk is at least three times higher for youngsters than for older people. Therefore, it is incontestable that a young country would count, ceteris paribus, more people killed on the roads than an old country. On the other hand, the number of roundabouts is supposed to be endogenous to road safety policy since this feature comes from a strategic arbitration between several features able to improve the safety of intersections.
It is obvious that international comparison of fatality rates alone is not satisfactory to evaluate performance since the calculation of the rates refers to a single denominator at the same time, which plays the role of the exogenous factor (vehicle fleet, kilometrage, population). This unique factor becomes the standardisation. It is a real step forward compared with the absolute number of fatalities, but remains insufficient because other important exogenous factors are not taken into account. And anyway, the fatality rate is either not available, or inaccurate or inconsistent since the estimation methods of kilometrage differ between countries.
If it is not entirely right to compare countries by using fatality rates, it is frequent to propose comparisons by using their progression which better reveals the safety improvements assuming constant exogenous factors over time. Unfortunately, in some cases, this assumption can not hold true. In other cases time series comparisons are dependant on the initial level of kilometrage and on the origin and end periods of comparisons (Andreassen, 1991).
It is therefore necessary to build more relevant or less defective indicators, which is the main objective of this paper (Section 4). This job required a statistical model that combines annual cross-sectional and time series data in 21 OECD countries from 1980 up to 1994.
In Section 2, I present some of the international data available in the International Road Traffic and Accident Database (IRTAD) and then recall, in the second section, the main lessons to be kept in mind from a literature survey in selecting the methods to be used to determine the new indicators of international comparisons. Doing so, the method will also reveal a property: the quantification of the effects of some exogenous factors on the fatalities.
All the factors have not been identified and the systematic statistical representations4 of these factors were also not found in some cases.
Section snippets
International data
The IRTAD (the OECD database) is one of the more convenient data providers for administration or research purposes since it offers a large range of time series data with regards to area of state, population with breakdowns by age, vehicle fleet (moped and mofas, motorcycles, personal cars and station wagons, trucks, buses and coaches), network length (urban, rural, motorways, A-level roads), kilometrage (urban, rural, motorways, A-level roads), kilometrage with breakdown by road users, injury
Epidemiological aspects of international comparisons
Descriptive statistics from IRTAD showed the great variability of safety and traffic indicators values between countries and over time. Descriptive epidemiology usually holds a more reliable mortality indicator than fatality rate: the age-standardised mortality rate which is curiously missing in the international reports or databases except in WHO's. The standardisation corrects the mortality by accounting for the age pyramid. If youngsters present a high risk all over the world, a young
An indicator for international comparisons
Looking at this review, the number of fatalities was modelled macroscopically with a pooling method combining time series and cross sections.
Discussion and conclusion
In Section 2 of this paper, I presented some of the international sources and data suppliers of road safety statistics. About ten institutions release international reports and/or supply databases about road safety. Statistics are usually available in the n+2 year except for general statistics such as injury accidents or fatalities available sometimes very quickly for immediate analysis (European Commission, 1994, OECD, 1997). Some of the statistics are released with a breakdown by road
Acknowledgements
This paper was written in 1997 while I was working for the French Road Safety Department (RSD). I wish to thank Olivier Noël (RSD) and all IRTAD Operational Committee members who encouraged me to carry out this international comparison about road mortality, especially Frank Poppe (NL), Peter Holló (H), Göran Nilsson (S) and Ekkehard Brühning (D). I am particularly indebted on Sylvain Lassarre (F) for his very precious advice and to two anonymous referees for their helpful suggestions. However,
References (51)
Population and registered vehicle data vs. road deaths
Accident Analysis & Prevention
(1991)Forecasting road accident casualties in Great Britain
Accident Analysis & Prevention
(1991)- et al.
An aggregate accident model based on pooled, regional time series data
Accident Analysis & Prevention
(1991) - et al.
Measuring the contribution of randomness, exposure, weather, and daylight to the variation in road accident counts
Accident Analysis & Prevention
(1995) - et al.
Estimation of linear models with crossed-error structure
Journal of Econometrics
(1974) - et al.
A critical review of macro models for road accidents
Accident Analysis & Prevention
(1991) Regression analysis of aggregate accident data: some methodological considerations and practical experiences
Accident Analysis & Prevention
(1986)Critique of Sam Peltzman's study
Accident Analysis & Prevention
(1976)The relation between motor vehicle accident deaths and economic activity
Accident Analysis & Prevention
(1984)Macroscopic models for traffic and traffic safety
Accident Analysis & Prevention
(1989)
Effects of macroeconomic conditions on the incidence of motor vehicle accidents
Accident Analysis & Prevention
Models explaining motor vehicle death rates in the united states
Accident Analysis & Prevention
Determinants of motor vehicle deaths in the United States: a cross-sectional analysis
Accident Analysis & Prevention
Motor vehicle inspection and motor vehicle accident mortality
American Journal of Public Health
Analysis of Longitudinal Data
Cited by (0)
- 1
IRTAD Web site, 1998. http://www.oecd.org//dsti/sti/transpor/road/stats/IRTAD