CONTENTS INTRODUCTION I. FRAMING THE QUESTION II. DO THE COMPREHENSIVE TAX BASE AND TAX EXPENDITURES LITERATURES ANSWER THE QUESTION? A. The Comprehensive Tax Base B. Tax Expenditures III. THE INTEGRATION QUESTION AS ORGANIZATIONAL THEORY: COORDINATION AND SPECIALIZATION IN PRODUCTION A. The General Problem of Divisions B. Departmentalization in a Governmental Context IV. APPLICATIONS: THE EARNED INCOME TAX CREDIT AND THE FOOD STAMP PROGRAM A. Background B. Food Stamps 1. Measurement Criteria 2. Error Patterns 3. In-Kind Provision of Benefits 4. Measurement Periods and Responsiveness a. Emergency Responsiveness Appears To Be Moderately Important to the FSP, but Not Central b. Other Programs, as Currently Constituted, Could Not Adequately Replace Food Stamps c. The Tax System Could Not Easily Be Made Responsive to Short-Term Food Needs d. Conclusion on Responsiveness 5. Conclusion C. The Earned Income Tax Credit D. Summary and Comments on the Relationship to Overall Reform CONCLUSION INTRODUCTION
This Article provides a new theory of tax expenditures. (1) Its argument is that the decision to implement a "nontax" program through the "tax system" (2) has little or nothing to do with tax policy. Instead, the tax expenditure decision, which we will also call the integration decision or the decision to combine tax and spending programs, is solely a matter of institutional design. It is about assigning projects such as tax collection, education, defense, or housing to specific units of government. Different groupings of activities will perform differently, and we should use those groupings that yield the best possible performance. The problem is similar to the problem of splitting up a corporation into divisions.
Suppose, for example, that we are considering whether an education subsidy should be implemented through the tax system or through a direct expenditure. The government might use a tax exclusion, deduction, or credit implemented by the IRS or, alternatively, it might use a direct grant implemented by the Department of Education. (3)
The two leading theories that purport to address this question of how to allocate government largesse both focus on tax policy. The most widely accepted theory, the comprehensive tax base theory, argues that a broad tax base distorts economic decisionmaking less than a narrow base and is also much simpler to administer. (4) To ensure that the tax base is as broad as possible, this theory suggests that spending and regulatory programs should not be implemented through the tax system; instead, they should be assigned to other agencies or departments. Integrating a spending program into the tax system, for example through a deduction, exclusion, or credit for some special activity, narrows the tax base and makes the tax system more complex. The standard or default response according to a comprehensive tax base theory, therefore, is that government spending programs, such as the education subsidy in our example above, should not be implemented through the tax system.
The other leading theory, the theory of tax expenditures, focuses partially on institutional design, but it, too, is ultimately a theory of taxation. (5) The key insight of this theory is to recognize the functional equivalence of putting a program in the tax system or somewhere else. While this insight relates to institutional design, the theory ultimately falls back on tax policy for recommendations. For example, the distributional effects of a policy are said to depend on whether it is correctly considered part of the tax base. The tax expenditure theory, just like the comprehensive tax base theory, would conclude that the education subsidy should not be implemented through the tax system.
In contrast to these theories, which focus on taxation, our theory focuses on institutional design--the question of how the government chooses to compartmentalize its functions. It is entirely irrelevant whether some piece of government policy complies with independent tax norms. (6) If the underlying policy is held constant, there are no effects of putting a program into or taking a program out of the tax system even if doing so hurts or enhances traditional notions of tax policy. Welfare is the same regardless of whether the program is formally part of the tax system or is located somewhere else in the government. If we mistakenly look only at the tax system instead of overall government policy, we will draw the wrong conclusions. Putting a program into the tax system makes the tax system look more complicated, but there is unseen simplification elsewhere. The tax system will seem less efficient, but the efficiency of government policy is unchanged.
The institutional design question is about dividing the government into units that will provide the best possible set of public policies and government services. Different groupings of government services will perform differently. Consider, for example, a proposal to have the IRS run the country's defense system, replacing the Department of Defense. The proposal is not as silly as it sounds. It would not mean that bespectacled revenue agents would be parachuting into the Hindu Kush wearing night goggles, camouflage, and pocket protectors. Instead, an intelligent Commissioner of Internal Revenue would allow his employees to specialize. Revenue agents would specialize in reading financial statements and soldiers would specialize in fighting. Policies under such a proposal might very well continue much as they do today. The reason why the proposal is not a good idea is that there are no benefits to coordinating these two functions of tax administration and defense. Moreover, there would be additional costs because at the very top level, where functions must be combined, administrators would be unable to specialize in these relatively distinct functions.
Consider instead a proposal to implement all federal welfare-type programs through the IRS. Proposals of this sort have been made frequently, often under the rubric of a negative income tax. (7) This may make sense, if there are benefits to putting welfare and tax into the same organizational unit: Both programs rely on income or wealth measurement, both need large-scale information and financial processing, and both have substantive policies, such as the policy of redistribution, that overlap significantly. The two programs thus might benefit significantly from coordination by a single agency.
The key variables from this perspective have nothing to do with tax policy. Instead, they have to do with the benefits of coordination between and specialization within various types of activities performed by the government. The Department of Defense needs highly specialized operatives, and thus benefits little from coordination with the revenue collection function. Welfare programs, on the other hand, may gain much from coordination with tax collection, and there may be low costs to losing the utility of separate units that can specialize in each function. The question is one of tradeoffs between the benefits of specialization on the one hand and the benefits of coordination on the other.
This intuition strongly contrasts with the usual tax arguments. For example, the Flat Tax is an attempt to provide a comprehensive consumption tax base. All the extraneous, nontax elements of current tax law would be removed. (8) The Flat Tax is said to be very simple, and it may be if one looks only at the tax system. But limiting the tax system to this one measurement would force other government programs to take up the slack: Programs of all sorts that are now embedded in the tax system will have to be implemented by other government agencies. Viewing the Flat Tax (or any comprehensive tax base) as simple requires ignoring the rest of government, relegating the complexity and mess of government spending and regulation to somebody else's backyard. There is no reason to believe, however, that the tax collection function should necessarily be separated from other functions of government--and there may be good reasons to believe that it should not be. The same is true for virtually all proposed fundamental tax reforms.
This Article expands on these intuitions. Part I frames the question as one of institutional design rather than tax policy. Part II discusses the comprehensive tax base and tax expenditures literatures. We argue that neither literature provides a convincing answer to the question of how government spending programs should be organized. The comprehensive tax base argument is the more prevalent of the two, but it ignores the basic problem of organization: It takes a completely tax-centric view of government and, therefore, leads to faulty conclusions. The tax expenditures literature addresses the problem--at least to some extent--as one of institutional design, but its analysis is neither complete nor convincing.
Part III approaches the problem from an institutional design standpoint. The study of organizations is old and deep, extending into sociology, economics, political science, and even anthropology and psychology. Covering even a small portion of this literature is well beyond the scope of one paper, but the problem can be divided into three more manageable pieces. First, we can view organizations as devices for coordinating specialized functions and, in particular, for separating production processes into tasks or divisions in the most efficient manner possible. Second, we can view organizations as a way to solve agency problems. Third, we can view the design of public organizations as a method of resolving public choice problems. The volume of literature in each of these areas is very large, and the more informal literature generally mixes these areas together. The portions that relate directly to the problem of divisions, however, are reasonably manageable and in some cases quite sparse. In this Article we address only the...