>For example, very heavy fines
>and penalties for drunk driving do reduce (but not
>eliminate) drinking and driving. One can argue that
>such penalties are regressive, unfairly penalize
>people who have no alternative modes of
>transportation, not 100% effective and so on - but it
>is diffciult to argue that their severity does not act
>as behavior modification.
Drunk driving is a result of impaired decision making. Driving to work or to the soccer game is just getting around. What's the basis for comparison here?
Besides, it's not clear that what you say here is true even for drunk driving (footnotes with sources at the site):
Measures of little or no value:
Incarceration. Jail or prison sentences for alcohol offenses, in spite of their great popularity, appear to be of little value in deterring high BAC drivers. In short, It appears that we can't "jail our way out of the problem." The perception of swift and certain punishment is more important than severity. Large fines appear to have little deterrent effect, according to research. Increasing the cost of alcohol with increased taxation would have virtually no impact on reducing drunk driving. Both research and common sense suggest that heavy drinkers are not deterred by cost and most minors who drink don't pay for or purchase their beverages.
http://www2.potsdam.edu/hansondj/DrinkingAndDriving.html