Traffic accidents, deaths and alcohol consumption

Abstract

The main goal of this article is to analyse the relationship existing among prices of alcoholic beverages, alcohol consumption and traffic fatalities for the Spanish Autonomous Communities during the time interval 1998 to 2002. Among the main results, we highlight a positive correlation between alcohol consumption and traffic mortality rate. Basically, governments implement two kinds of policies to reduce the traffic mortality rate. One is oriented to control the supply of alcohol by increasing alcohol taxes. The other is oriented to preserve traffic security, increasing the number of sanctions for traffic rule infraction. We find evidence that both policies exert a positive influence in the reduction of traffic fatalities. There is no empirical evidence to indicate that being a novice driver increases the tendency to be involved in a mortal traffic accident.

Acknowledgements

The authors are indebted to the anonymous referee and Donald S. Kenkel for their valuable comments in previous versions of this article. An earlier draft has been published as a working paper by FUNCAS; so the authors appreciate the recommendations received by this institution. Finally, the authors would like to express their thanks for the financial support provided by the Spanish Ministry of Education. The usual disclaimer applies.

Notes

¹ In all these statistical numbers, a distinction could not be made between traffic fatalities caused by the driver's own fault (e.g. death of drunk driver) or those caused by someone else's fault (e.g. drunk drivers kills other person). From the report of the Spanish General Directorate of Traffic (2005), we conclude that to analyse drunk driving, it is important to study independently the number of drivers who drove under the influence of alcohol and the number of people who died as a consequence of having been involved in a traffic accident, irrespective of whether the person who died caused the traffic accident. Therefore, unless the contrary is said, the number of victims refers to the total number of people who died as a consequence of having been involved in a traffic accident in which the driver presented a BAC higher than that permitted.

² Wine products have been exempt from special indirect taxes to protect the competitiveness of European wine industry. However, the European Commission has been debating the adequacy of this exemption.

³ These papers have analyzed the traffic accidents in United States, where the fixed spatial effects are related to the different states.

⁴ The number of deaths by traffic accidents reflects the total number of people who died as a consequence of having been involved in a traffic accident independent of whether they provoked it or not.

⁵ The modelling of traffic accidents as a function of alcohol consumption, Equation 3, is not exempt of weaknesses. For instance, whether alcohol consumption increases the probability of a traffic accident depends on: when the alcohol is consumed (just before driving); where the alcohol is consumed (e.g. at home or away from home, with dinner) and how much alcohol is consumed in a short period of time. So two people, or two regions or two countries, can have the same amount of alcohol consumption but very different levels of traffic risk. Regularly drinking a glass of wine with dinner may not be a problem; drinking seven glasses of beer on a Friday night could cause a drunk driving accident.

⁶ The mortality rate of traffic fatalities is the best empirical approximation to analyze the relationship between alcohol consumption and traffic fatalities (Young and Likens, 2000; Young and Bielinska-Kwapisz, 2006). Although all traffic accidents are not the result of drunk driving, there is a strong correlation between both variables.

⁷ Other useful specification lies in the Linear Regression Models. Young and Likens (2000) compared the results of both specifications (linear and logistic) and found just small differences. We have also carried out estimations from both specifications, concluding that the results are very similar. Results are available on request.

⁸ Both assumptions are recurrent in the literature; see for example the articles of Mast et al. (1999), Young and Likens (2000) and Young and Bielinska-Kwapisz (2006).

⁹ The Augmented Test Regression and Hausman Test are included in the software package Stata 9.0. More information about these tests is available at: http://www.stata.com/support/faqs/stat/endogeneity.html

¹⁰ To check the presence of heteroskedasticity, we carry out the test of White and the test of Breuch–Pagan/Cook–Weisburg, which are also available in the software package Stata 9.0. Both test results are algebraically identical and are available from the authors upon request.

¹¹ The temporal interdependency of alcohol consumption derived from the Addiction Theory (Becker and Murphy, 1988) postpones the effect of public policies; thus, the individual requires a period of time to adjust to the new changes in the parameters that define alcohol consumption (McKenzie, 1991).

¹² We have recurred also to the Gross National Product (GNP) with which we have obtained similar estimations to that described in the present article.

¹³ The variable related to traffic sanctions informs us of the number of sanctions expedited by the Policy as a consequence of traffic controls.

¹⁴ Novice drivers are not allowed to drive with BAC higher than 0.3 for the first 2 years (Royal Decree 1428/2003 of 21 November).

¹⁵ Alcohol taxes are fully passed through to final prices (Kenkel, 2005). Taking into account that the Spanish tax rate applied to beer is around 16.75% of the final price (including all indirect taxes, that is VAT and special tax), an increase by 10% of the beer taxes will increase the final price of the beer by more or less 9.4% (Law 37/1992 of 29 December and Law 38/1992 of 28 December) (Table 6).