The Impact of State Taxes on Self-Insurance: Closing remarks

5. Closing remarks
In summary, this paper finds self-insurance of property-casualty risks increases in state taxes. Tests are conducted assessing the relation between a state’s property-casualty insured losses and its tax levy on the insurance industry. As expected, a negative relation holds for nonautomobile coverage and automobile physical damage coverage. Similar relations are detected for workers’ compensation benefit payments. These findings are consistent with consumers opting to self-insure rather than bear the incidence of higher insurer taxes.
We fail to detect a negative relation for liability coverage, consistent with self-insurance not being an option to increased taxes when risk retention is too costly. In most states, selfinsurance of liability risks results in the forfeiture of the right to drive. The inelasticity of the demand for liability coverage results in consumers bearing its tax incidence. In a more positive light, few deadweight costs should arise from taxing automobile liability coverage.
To our knowledge, this is the first study to link state taxes and insurance coverage in a state. It should encourage additional investigations of the impact of subnational taxes on business activity. Such studies are warranted because U.S. businesses pay more in state and local taxes than in Federal taxes and the spread is widening as devolution increases state budgetary pressure. From a policy perspective, the findings suggest that suboptimal risk sharing results from taxing insurance products that have elastic demand. Consequently, states with lower insurer taxes (and accordingly less self-insurance) should fare better during periods of catastrophic damage than high-tax states, ceteris paribus. Additional analysis, however, is needed to assess the social welfare implication of insurer state taxes. It is beyond the scope of this study to determine whether the revenue raised from taxing insurers offsets the risk-sharing suboptimalities.
Finally, readers should interpret the inferences drawn from this study with caution for at least two reasons. One, ex-ante expectations about the coverage provided in a policy is imperfectly measured by ex-post insured losses. Two and perhaps more importantly, because the theory is not rich enough to specify the interaction among states, insurers, and consumers, the paper relies on several simplifying assumptions. They include: (a) unilateral responses by insurers, governments, and consumers; (b) exogenous state tax policy; and (c) no behavioral responses to self-insurance.