Introduction Since the mid-1970s much theoretical discussion has occurred on the role of investment income in property-liability ratemaking. Often cited in this regard are the articles by Quirin and Waters (1975), Biger and Kahane (1978), Fairley (1979), Hill (1979), and the follow-up advances by Myers and Cohn (1987), Hill and Modigliani (1987), and Kraus and Ross (1982). Surveys of the general developments of this period can be found in Cummins (1990a, 1990b, 1991) and Derrig (1990). Interest in this subject is not restricted to academicians. Actuarial notice of the topic resulted in a full Call Paper Program on total returns for insurers at the 1979 meeting of the Casualty Actuarial Society. Applications of the theoretical models, primarily those derived from the capital asset pricing model (CAPM), were made for both the automobile and workers' compensation lines of property-liability insurance in several states, most notably in Massachusetts. Derrig (1987) provides a developmental history of the important issues in Massachusetts for the period 1976-1984. Some of the empirical results of the Massachusetts experiment through 1990 can be found in Derrig (1993). Regulators, aware of the pressures of inflation on costs and the availability of relatively high current market yields for investments in the late 1970s and early 1980s, began to pay close attention to the question of the inclusion of investment income in ratemaking (a survey of the early methods is provided in Williams, 1983). Explicit recognition of investment income in ratemaking also emerged in the prior approval states of Florida, Minnesota, North Carolina, and Texas. Finally, the National Association of Insurance Commissioners (NAIC) undertook an extensive review of the question, which prompted recommendations by an industry advisory committee (Haayen, 1983). Throughout this period, little, if any, attention was paid to the crucial role that federal income taxes would play in any ratemaking model. The simple no-tax, no-risk, one-period result of an underwriting profit provision equal to minus the risk-free rate appeared in both the Quirin and Waters and Biger and Kahane articles. Their results led to the surprising observation that the fair premium, or underwriting profit provision, was independent of the amount of supporting surplus (Biger and Kahane, p. 124), but that actual results do seem to depend upon a solvency constraint (Quirin and Waters, pp. 438-439).(1) The potential importance of the omission of both risk and taxes was recognized immediately in Brennan's (1975) comments on the Quirin and Waters paper. On taxes, Brennan noted: I think that one still might find that equilibrium expected underwriting losses were less than predicted by the QW model which focuses on the similarity of insurance companies to investment companies while neglecting a significant difference, namely that insurance companies, must pay taxes on their investment income. So, in opting to become an insurance company rather than an investment company a significant tax burden is accepted. Equilibrium considerations would suggest then that it is only reasonable to expect some countervailing advantage in the form of the ability to borrow at preferential rates from policyholders (pp. 446-447). Hill (1979) also observed the independence of the profit margin from surplus requirements but went on to derive a simple but necessary dependence in a capital asset pricing model framework with corporate taxes on investment income.(2) Fairley's (1979) CAPM profit margin equation with taxes essentially recognizes the same phenomenon. Fairley begins to reveal the potentially large effects from the estimation of an appropriate tax rate. He estimates a three percent increase in the required profit margin when the model insurer is assumed to invest only in U.S. government bonds, taxable at the full 1979 corporate tax rate of 46 percent, rather than investing in a tax-preferred diversified portfolio, taxable when combined with underwriting income at an average rate of about 20 percent. …