This article develops a set of models for the determinants of automobile fatalities with particular attention devoted to the effects of increased cell phone usage. Cell phones have been associated with both life taking and life-saving properties. However, prior statistical evaluations of the effects of cell phones have led to fragile results. We develop in this article econometric models using time-series data, allowing for polynomial structures of the regressors. The models are evaluated with a set of specification error tests providing reliable estimates of the effects of the various policy and driving-related variables evaluated. The statistical results indicate the effect of cell phones is nonmonotonic depending on the volume of phones in use, first having a net life-taking effect, then a net life-saving effect, followed finally with a net life-taking effect as the volume of phone use increases. (Author/publisher).
Abstract