January 15, 2002


Issues and Options for States
by Michael Mazerov

I. Introduction

PDF of the full report
The Rising Regressivity of State Taxes:
- HTML of the press release
- PDF of full report

If you cannot access the files through the links, right-click on the underlined text, click "Save Link As," download to your directory, and open the document in Adobe Acrobat Reader.

During the late 1990s, many states enacted large tax cuts as they enjoyed their best fiscal health in years. Now, most states are in financial distress because of the recession that began in early 2001, and some are likely to find it necessary to raise taxes — at least temporarily. Whether states are considering tax cuts or tax increases, however, two questions are important and often asked by policymakers, the media, and the public: "How much will this tax change cost?" and, "Who will pay more or less in taxes as a result of this change?"

All states have developed sophisticated methods for determining how much proposed tax cuts will cost or tax increases will raise. In addition, they have methods in place for estimating the total amount of revenue that will be generated by their current tax structure.

States are much less well-prepared to answer the second question: "Who will pay more or less in taxes as a result of this change?" Fewer than one-fifth of the states have developed the capacity to analyze comprehensively how proposed changes in their tax laws would affect the amount of taxes owed by different income groups in their populations or how total tax obligations are distributed across income groups at a particular point in time. Even fewer states — only Maine, Minnesota, and Texas — actually require that a distributional analysis of their tax laws, formally called a "tax incidence study," be conducted.

At the federal level, by contrast, both the Treasury Department and Congress’s Joint Committee on Taxation provide at least some information on how major tax proposals will affect taxpayers at a variety of economic levels. The information contained in these "distributional analyses" are frequently cited during debates and often affect the contents of final tax packages. This type of information, however, rarely is available as states debate tax policy.

State Tax Systems and Local Property Taxes

This report presumes that an ability to analyze the distribution of liability for local property taxes among households of different income levels is an essential element of a comprehensive "state" tax distribution analysis capability. Although the vast majority of property taxes are levied and collected by school systems, cities, counties, and other local governments, the level, structure, and distribution among households of local property taxes generally is tightly constrained by state law. Moreover, a large share of the attention devoted by state legislatures to tax policy in recent years has involved efforts to relieve and change the distribution of property tax liabilities by increasing state aid to local governments financed through higher state income or sales taxes. When income or sales taxes are increased for this purpose, the distributional impacts of these two state taxes are sometimes changed as well. Thus, although this report will refer to the distributional impact of state tax systems or state tax changes, it should be understood that these are intended to encompass as well changes in the distribution of local property taxes.

The benefits to a state and its residents of developing the capacity to determine the incidence of its tax structure are many.

There are no insurmountable technical or logistical problems to producing such information. A few states do provide regular, comprehensive "tax incidence" studies that tell legislators and the public how the state tax burden is distributed among the population and how specific proposals would change that burden. Such analyses require both the technical capacity to conduct such analyses and the procedural requirements that the analyses be conducted and made publicly available in a timely fashion during legislative debates.

Experience shows that tax incidence analyses can inform and improve the debate over tax proposals. For example, in Minnesota in 1997, the governor proposed the creation of tax-free education savings accounts. A Department of Revenue analysis showed that the bulk of the benefits from such a proposal would go to families with incomes above $100,000, an analysis that arguably prompted the legislature to reject that proposal and instead enact tax breaks for education that were more tightly targeted to low- and middle-income taxpayers. Similarly, an analysis by the Legislative Budget Board in Texas in 1997 showed that a sweeping proposal submitted by the governor to cut property taxes and raise some sales and business taxes would have provided the largest benefits, both in dollar and percentage terms, to the wealthiest Texans. The proposal did not pass.

This report describes in detail the primary methods states can use to assess the distributional impact of their tax structures on families or households. Three methods are described. They are called, respectively, the economic incidence model, the initial tax impact model, and the representative taxpayer model.

Both the economic incidence model and the initial tax impact model are based on developing a sample of actual taxpayers that is representative of the total population of taxpayers in the state. These types of models yield data on the total and average amount of taxes paid by taxpayers in different economic circumstances. The economic incidence model is more comprehensive in that it incorporates assumptions that certain taxes initially imposed on business — such as corporate income taxes and property and sales taxes paid by businesses — are passed through or shifted to households in the form of higher prices, lower wages, or reduced returns on business investments. The initial tax impact model, by contrast, generally includes only those taxes that are paid directly by individuals and households — such as personal income taxes, sales taxes, and property taxes on owner-occupied homes.

The representative taxpayer model is a very different method of determining the impact of a state’s tax structure on persons in different economic circumstances. This method involves constructing profiles of a limited number of hypothetical "typical" families and determining their tax liability under the existing state and local tax structures or under different tax proposals.

For each method, the report includes a description of the method, an example of the use of this type of model, a discussion of the pros and cons of the method, and case studies of its use in state legislative debates. In addition, the report includes a summary of the existing capacity of the states to do these types of analyses, and a discussion of some of the choices states must make in setting up this capacity.

The information and examples included in this report can assist states in developing or expanding a much-needed capacity to analyze the distribution of their tax systems. In this era of computerization and information access, no state should determine tax policy without the ability to assess the impact of that policy on the state’s citizens at all income levels.

Evaluation Criteria for State Tax Incidence Models

To be useful, a tax incidence model must

  • provide a comprehensive picture of the distribution of state and local tax obligations,
  • provide results that are readily interpretable,
  • allow rapid testing of the distributional impact of a wide variety of alternative tax policies,
  • be robust (that is, provide results that do not change dramatically if the underlying assumptions in the model are varied to a modest degree),
  • be based upon commonly-accepted economic assumptions,
  • be as inexpensive to build and maintain as possible, and
  • use readily-available and reliable data.

Of course, in seeking to satisfy these criteria, significant tradeoffs are likely to be encountered. For example, the more comprehensive a model, the higher its construction and maintenance costs are likely to be.

The three principal approaches to tax distribution analysis differ significantly with respect to their satisfaction of these criteria. Economic incidence models unquestionably provide the most comprehensive picture of the distribution of state and local tax liabilities among different income groups. They tend to be constructed in a manner that permits analysis of a wide variety of quite specific changes in a state’s tax structure. However, they are more costly to build and maintain than initial tax impact and representative taxpayer models. Moreover, in their effort to account for taxes imposed on businesses but passed through to individuals, economic incidence models must rely on economic assumptions subject to dispute and on economic and tax data of less-than-ideal quality. When they are used to analyze the overall incidence of a state’s entire tax system, economic incidence models are reasonably robust. However, when they are used to analyze the distributional impact of changing a single tax — particularly a tax imposed on businesses — changes in economic incidence assumptions can have a significant impact on the output of the model.

Initial tax impact models inherently give a less complete picture of the distribution of tax liabilities than do economic incidence models because they do not seek to evaluate the incidence of business taxes. This can be a significant shortcoming, particularly when it comes to analyzing the incidence of proposed legislation that simultaneously changes household and business taxes. On the other hand, initial tax impact models have fewer data reliability issues than do economic incidence models and, in foregoing business tax analysis, avoid many economic theory-related controversies about tax incidence.

Representative taxpayer models can provide only a very limited illustration of the distributional impact of a state’s tax system or of proposed changes in tax laws. The output of such models cannot be generalized to the entire population of a state nor summarize the distributional impact of a state’s tax system. Moreover, representative tax models are not robust; small changes in the underlying assumptions about the tax-related characteristics of the hypothetical "typical" families in the model can have a significant impact on the models’ output. In addition, such models generally do not incorporate detailed provisions of state tax structures, limiting their usefulness in analyzing the distributional impact of many proposed tax changes that policymakers may consider. (For example, a representative taxpayer model typically would not permit an analysis of the incidence of extending the sales tax to a few specific services.) Notwithstanding these limitations, representative taxpayer models can play a useful role in tax incidence analysis. They have the potential to illustrate the distributional impact of specific changes in tax law in a manner that is transparent and easy for policymakers to understand. Moreover, they can be constructed using off-the-shelf personal computer spreadsheet software, making them relatively simple and inexpensive to build and maintain.

Click here for PDF version of the full report.