Abstract
The purpose of this paper is to estimate the effect of the preliminary breath test law on drunk driving in Canada. Using an Annual time-series of provincial cross-sectional data consisting of six provinces from 1952 to 1988, it is found that the law seems to have a negative effect on non-alcohol-specific fatality and fatal accident rates. However, the estimate of the effect is sensitive to model pecifications, perhaps due to high multicollinearity among variables. But the law seems to have a positive effect on the alcohol-specific fatal accident rate using data from the province of Ontario for the period 1954–88. Our interpretation is that the breath test law has two simultaneous effects on drunk driving: deterrence and detection effects. The results suggest that the deterrence effect is weak and the detection effect is dominating. One robust result is that the real beer price has a negative effect on drunk driving traffic accidents.