Relationship between fatality of drunk driving and related punishment has long been an interesting issue for researchers. Many outstanding and remarkable researches about effects of policies on traffic fatality have been conducted. However, current studies on this field offered confusing and opposite conclusions. Dee (2001) applied an OLS regression model with dummy variables representing state and time effects to explore the relationship between fatality and BAC limits. His study found out that 0.08-BAC limit did reduce traffic fatality related to drunk driving by use of panel data from the 48 states for years 1982-1998. His estimation of the effects of 0.08-BAC limit on traffic fatality was around 7.3 percent with minus sign.
He further pointed out that the effect would be reduced to 5.3 percent if BAC limit was raised to 0.10. Other studies which focused on the relationship between driver BAC levels and traffic crash risk also provided conclusions supporting Dee’s. Zador et al. (2000) examined the relationship between BAC levels ranging from 0.08 to 0.10, and related traffic risk which was quantified from 6 to 24 in different groups. Observations were categorized to these groups depending on their age and gender. Zador et al. found out that drunk driving in fact affected ability of driving, such as reaction time when facing emergency. Further research was conducted by Eisenberg.
He (Eisenberg 2003) constructed a DID model (Difference in Difference) and found out that marginal effect of raising BAC limit from 0.08 to 0.10 was around 2 percent which was consistent with Dee’s conclusion. Additionally, lowering BAC limits seemed to have delayed effects on traffic fatality. According to Eisenberg, there were six years gap between the year when BAC limit was lowered and fatality fell. The most important findings from Eisenberg’s study is that graduated license laws had been playing an important role in reducing fatality rate.
However, other studies proposed contrast conclusions on this field. Gao (1999) found out that reduction of traffic fatality in fact could not be attributed to law of BAC limit. He claimed that other policies should be introduced in order to maker law of BAC limit effective. Likewise, Freeman (2007) also pointed out that there was no significant evidence to support the Dee and Eisenbergs’ conclusions. On one hand, by means of regressing a conventional model with extended sample, Freeman could not find out the effect proposed by previous researches. On the other hand, he indeed found out that introduction of other policies, such as ALR law and seat belt laws, had consistent influence on reducing traffic fatality.
There are a lot of reasons to explain confusing and mixed conclusions on this field. Sample size and correlated disturbances are two of potential reasons. Additionally, previous researches are unable to distinguish traffic fatality rate and fatality rate directly related to drunk driving.
