Limiting Tax Expenditures
The federal government devotes over a trillion dollars each year to tax provisions that pursue “nontax” goals. Scaling back these tax expenditures should be a high priority. Yet one-size-fits-all limits are often proposed, and are not good policy. Each tax expenditure generates its own mix of positive externalities and private benefits (or “programmatic benefits”). To choose the right limit, we should consider what programmatic benefits we would lose. The goal should be to reap programmatic benefits at lower cost. Different strategies are appropriate for each tax expenditure, including: tightening the definition of favored conduct; focusing on claimants who are easiest to motivate; favoring claimants who use the subsidy more effectively; calibrating how much favored activity we subsidize; and changing the government agency that administers the subsidy. We also should account for excess burden and distribution. Does repeal or a limit influence labor or savings decisions? Does it affect planning and administrative costs? Does it bring is closer to the distribution we want?
In addition to proposing this three-part framework for limiting tax expenditures, which focuses on programmatic benefits, excess burden, and distribution, this Article also analyzes seven different limits. They have very different effects. For example, a “cap” eliminates the subsidy for high levels of favored activity. In contrast, a “floor” disallows the subsidy for low levels. “Haircuts,” “maximum fractions,” and “phaseouts” preserve the subsidy for both high and low levels of favored activity, but in weakened form. Each limit offers a different mix of strengths and weaknesses, making it a better fit for some tax expenditures than others.
Like limits, tax expenditures also vary in systematic ways. This Article identifies an important distinction among them. For some tax expenditures, marginal benefits vary only with the activity level of all claimants in the aggregate; for others, marginal benefits also vary with the activity level of each claimant. When we subsidize green energy, for instance, the aggregate is our main concern; the goal is to replace as much carbon-based energy as possible, and it matters less who is doing so (as long as they do it well enough). In contrast, when we subsidize health insurance, we care a lot about how much insurance each individual has. The difference between what this Article calls “aggregate” subsidies (like green energy) and “individually-based” subsidies (like health insurance) can influence the type of limit we want. For example, caps are likely to be a better fit for individually-based subsidies than aggregate ones, since we care more about how much each claimant claims.
This Article also makes a number of other recommendations, including: first, the subsidy rate often should vary for different tax expenditures; second, instead of using “basket limits” that govern a group of tax expenditures, we should tailor a separate limit for each one; and third, the subsidy rate often should vary with income.