The Impact of State Taxes on Self-Insurance: TAX coefficients

In addition, the TAX coefficients for the physical coverage regressions are significantly less than the TAX coefficients for the automobile liability coverage regressions. In fact, the liability TAX coefficients are always positive, though not significantly different from zero. These findings are consistent with consumers being insensitive to upward adjustments in the cost of liability insurance because, unlike physical insurance, liability coverage is a necessary condition to operate an automobile. Within the liability and physical groupings, the regression coefficient estimates on the control variables are stable across years. However, the coefficients are in the predicted direction and significant more often for the liability regressions. Population is always positive and highly significant for both liability and physical coverage. Wealth, population density, and thefts are always significantly positive in the liability regressions, but significantly different from zero in the physical regressions only once (density in 1993). Catastrophic damage never has explanatory power in either set of regressions, which is not surprising because catastrophes only account for a tiny fraction of all automobile claims.
As predicted, the regression coefficient estimates on COMPULSORY are always positive (significant at the 1 percent level) in the automobile liability regressions, indicating that requirements to carry liability insurance reduce self-insurance. As expected, the COMPULSORY coefficient is never significantly different from zero in the automobile physical regressions. Finally, additional tests segregate insured physical losses between commercial and personal property. Because businesses have a lower percentage of their wealth in automobiles and their automobiles are more widely dispersed, self-insurance should be less costly for them than for individuals. If so, business insurance should be more price elastic than individual insurance. Untabulated results are consistent with this prediction. Specifically, the TAX coefficients for commercial losses are at least twice the TAX coefficients for personal losses. The hypothesis that the TAX coefficients are identical for commercial and personal automobile insurance can be rejected at the 5 percent level in 1993 and the 10 percent level in 1994 and 1995.