Vanishing Corporate Income and Government Spending in Distribution Models

Abstract

Many distribution models that are workhorses for tax policy, and more broadly for research on immobility and inequality, are insufficient welfare measures of either the status quo or policy proposals. This paper explores two fundamental modeling problems: The omission of corporate income, and the open-model complication caused by incomplete distribution of government spending and deficit effects.Capital-gain-realization (CGR) based models understate corporate income, and therefore overstate progressivity, by failing to capture even the corporate net income tax base itself, not to mention the omission of assorted corporate tax expenditures (both of these omissions contrast with the CGR model’s income inclusivity with respect to the individual income and payroll taxes). Most damaging is the CGR model’s inability to capture deferral, which is the primary tax benefit accorded corporate (and other capital) investment. An inclusive corporate income method is offered as an alternative to both the CGR method and the full-accrual approach (the latter is theoretically desirable, but, as a recent attempt shows, its implementation is problematic), with implications for the status quo, proposals to restructure the corporate tax, and ongoing empirical research on the potential shifting of the burden of the corporate tax. Traditional models also generally omit the distribution of full government spending and deficit effects. These omissions wouldn’t be an issue if there were supplemental models that contemporaneously measured these effects, but that additional modeling is absent from policy discussions, with little warning about these omissions given to distribution model users. Omitting full distribution of government spending and deficit effects biases traditional distribution models in favor of tax cuts and deficit spending, and makes the overall system look more progressive than it really is. In light of recent CBO and other research, these omissions are fixable, or at the very least should be heavily caveated in distributional models.