NOFAULT is a categorical variable indicating whether the state is one of the 14 with nofault automobile coverage. Because Witt and Urrutia (1983) find the price of automobile insurance is less in states with no-fault insurance, a negative coefficient is expected for NOFAULT.

INSPOP is number of insurers per 1000 people in the state. Browne and Hoyt (1995) find that a 10 percent reduction in the number of property-casualty insurers results in an 82 percent reduction in the industry’s insolvency rate. They interpret this finding as evidence that fewer insurers result in less competition and higher prices. To the extent INSPOP captures competition, its coefficient is expected to be negative. itat on

Insurer-specific Control Variables

The remainder of the control variables addresses cross-insurer variation in the price of insurance. DIRWRITE is a categorical variable indicating whether the insurer directly sells its policies, as opposed to using independent agents. Several studies (e.g., Joskow 1973; Cummins and VanDerhei 1979; Cummins, et al. 1980; and Witt and Miller 1981) report the price of insurance decreases in the presence of direct writers. Accordingly, a negative coefficient on DIRWRITE is expected.

SIZE is the natural logarithm of the insurer’s admitted assets.9 Larger insurers can charge lower prices because they can amortize their fixed costs over a larger pool and diversify their risks more easily. Thus, PRICE is expected to be decreasing in SIZE.