Progressivity: A Shot in the Dark
Scott, Charles E., Derrick, Frederick W., Atlantic Economic Journal
With three out of five states in serious financial trouble, with total state deficits projected to be at least $35 billion in fiscal 1992, and with balanced budget restrictions in 49 out of the 50 states, states are in dire need of revenue enhancements [Thomas and Thomas, 1991]. Legislatures often view increasing progressivity as an option for generating revenue.
The progressivity-revenue relationship has been a source of deliberation by theorists for over a quarter of a century. The theoretical investigations have led to contradictory hypotheses concerning the progressivity-tax revenue relationship.(1) Buchanan [19641 hypothesized that progressive taxes produce higher revenues than comparable proportional taxes using an argument based on average cost shares. In contrast, work by Brennan  and by Brennan and Buchanan  present arguments based on marginal cost shares, supporting an inverse relationship between progressivity and tax receipts. However, when adjustments to equilibrium are, allowed, the theories relating progressivity to revenue lead to conflicting predictions.
With conflicting hypotheses found in the theoretical models, with the models' limitations over time, and with the significant adjustments in the federal and state income tax system as in the 1980s,(2) the exploration of the progressivity-revenue relationship over time is germane to policy. In the present study, pooled cross section and time series analysis is used in the context of a median voter model to address the empirical issue of revenue enhancement. Emphasis is on whether changing the progressivity of the state's income tax system can be a source of additional revenues.(3)
Prior progressivity-revenue studies are based on cross sectional models and have limitations when applied over time. One period models, with the exception of the fiscal illusion mode s, do not allow for changes occurring during the business cycle, including inflation and employment. Similarly, the studies do not account for changes in tax laws and the resulting individual taxpayer or voter responses. They assume static income distributions and ignore potential tax payer mobility. In addition, the one period models employed in the studies suggest contradictory hypotheses. Thus, the progressivity-revenue relationship over time is an empirical question which must allow for adjustments in the structure and the resulting counter adjustments by citizens and competing states.
I. Specification of Revenue Factors
Democratic elections and political decisions can be analyzed by considering the median voter since they vote with the majority. Thus, the specification of the progressivity-revenue relationship is structured to control over time for the incentives facing the median voter.(4) These include the tax price, the potential impact of fiscal illusion, and adjustments over the business cycle. Revenue, in this study, is defined as state income tax revenue per $1,000 of income in the state to control for differences in income across states.
The tax price is proxied in this model, as in the Buchanan-Brennan analysis, by the average income tax rate for the median income individual. While the evaluation of total receipts and tax prices from all revenue sources would be preferable, there is no consistent measure of either tax price or progressivity at this aggregated level. The income tax rate is used as a proxy for the tax price for measurement consistency with the income-based measures of progressivity and revenue.(5) In addition, the income tax is the most commonly used source of progressivity and the most visible of tax prices.(6) While common perception leads to the expectation of a positive relationship, the possibility of a Laffer curve relationship leads to an undefined expectation.
Progressivity is measured by a Gini coefficient, which measures the progressivity of the rate structure rather than tax payments. …