The Impact of State Taxes on Self-Insurance

The Impact of State Taxes on Self-Insurance1. Introduction
This paper assesses whether insurers state taxes reduce purchases of property-casualty insurance coverage in a state. Insurance rates should be higher in states with higher taxes, assuming the demand curve for insurance is not perfectly elastic. Consequently, consumers should self-insure more in those states. Self-insurance includes forgoing insurance completely, purchasing lower levels of coverage, or accepting higher deductibles. To test the proposition that self-insurance is increasing in state insurer taxes, all property-casualty insured losses in each state are regressed on insurers’ total taxes in that state. Publicly-available, annual accounting reports, which all insurers must file with state regulators, provide the data from which aggregated state insured losses are computed. Tests are conducted separately for 1993, 1994, and 1995.
The dependent variable is designed to capture the extent to which consumers shift property-casualty risks to insurance companies. Unfortunately, risks (or coverage or selfinsurance) are unobservable. One possible proxy for insurance coverage is premiums. We use insured losses, instead of premiums, to measure coverage because Petroni and Shackelford (1999) report that insurers understate (overstate) the premiums earned in states that apply a higher (lower) premium tax rate. Because the premium tax, which is the primary state tax, is levied on insurers’ gross receipts, insured losses do not affect the premium tax liability. Unlike losses, premiums are further biased because they are reported inclusive of the premium tax. Therefore, insured losses should provide a less biased measure of insurance coverage than premiums. To control for unanticipated losses, the regression’s explanatory variables include an estimate of catastrophic damage in the state. We also control for other potential sources of crossstate claim variation, such as population and wealth.
Besides the usual research limitations, such as measurement error and inadequate power, this paper may fail to detect a relation between self-insurance and taxes for at least two reasons. First, tax plans may fully eradicate cross-state tax variation. Slemrod (1996), Grubert and Slemrod (1996), Shackelford (1999), and Olhoft (1999), among others, model tax planning as continuing until the costs of tax planning exceed the taxes avoided. Slemrod (1990, 1992) constructs a hierarchy of tax avoidance where firms undertake relatively low-cost accounting and financial tax plans that exploit subtle legal distinctions to prevent taxes from affecting their production and marketing (“real” responses). Consistent with these models, extant studies document that property-casualty insurers make seemingly low-cost accounting and organizational choices that reduce their state taxes. If successful, these tax avoidance strategies may eliminate cross-state statutory tax differences and prevent taxes from differentially affecting pricing decisions. If so, state taxes should not differentially affect insurance rates or selfinsurance.