Unemployment and highway fatalities

We considered unemployment, an often overlooked covariate of highway fatalities, hypothesizing that (1) as unemployment rises, aggregate driving decreases, especially among the unemployed, and as driving decreases, fatalities should decrease; (2) unemployment may influence drinking--some among the unemployed may drink less due to lower incomes, while others may drink more due to stress so the net effect would be ambiguous; and (3) unemployment may increase aggregate levels of stress and unhappiness, which can result in poor concentration on driving and thus, in turn, should result in more accidents and deaths. We used data from fifty states and the District of Columbia from 1976-1980, representing 255 observations. (No prior study has as many observations or controls for as many covariates.) Using econometric models of the data, we present evidence for two of the three hypothesized effects of unemployment. We conclude that, if the number of miles driven is held constant, worsening unemployment leads to higher fatality rates, most likely due to stress effects. But because more unemployment means less driving, increases in unemployment, on balance, are associated with decreases in fatalities.