Document Type



Each year the federal individual income tax delivers over $500 billion worth of incentives intended to encourage socially beneficial activities. Despite their efficiency rationale and substantial price tag, relatively little attention has been paid to the question of what is the most efficient form for these tax incentives. Currently the vast majority operate through deductions or exclusions, which link the size of the tax benefit to the taxpayer's marginal tax bracket. This Article argues that uniform refundable credits are a more efficient approach for tax incentives intended to correct for positive externalities, absent evidence that externalities or elasticities associated with the subsidized activity vary by income class. This conclusion holds even if no positive externalities are present. Moreover, even when evidence of differences in externalities or elasticities does exist, the most efficient subsidy is almost certainly still some type of refundable credit. The efficiency benefits of refundable credits are further magnified by their tendency to automatically smooth household income and macroeconomic demand. This Article therefore proposes a dramatic change in how the government should provide tax incentives for socially valued activities: the default for all such tax incentives should be a uniform refundable tax credit.

This Article also considers the arguments of refundable credit opponents. It concludes that concerns based on administrative and compliance costs are generally overstated. In addition, it finds that the vast majority of those who benefit from current-law refundable credits have positive income tax liability over time. Refundable credits are thus frequently the “rough justice” equivalent of allowing carryovers and carrybacks, which are common in other parts of the tax code. The Article therefore argues that the objection that tax incentives should not be structured as refundable credits because all Americans should pay some income tax is not only conceptually problematic; it is generally empirically unpersuasive on its own terms.

Date of Authorship for this Version

October 2006