The Institute on Taxation & Economic Policy (ITEP) has engaged in research on tax issues since 1980, with a focus on the distributional and revenue consequences of both current law and proposed changes. ITEP’s timely, accessible and accurate analyses help inform policymakers, advocates, the media and the general public about the fairness, adequacy and sustainability of proposed changes to federal, state and local tax systems.
The ITEP microsimulation tax model is a tool for calculating tax revenue yield and incidence, by income group, of federal, state and local taxes. The ITEP model is frequently used to analyze federal and state tax proposals and to look at the impact of current tax policies on issues of public concern.
How does ITEP calculate sales tax incidence?
There are broadly two ways of thinking about sales tax incidence: the initial incidence (who makes the purchases that result in sales tax liability) and final incidence (who ultimately bears the cost of the tax). ITEP’s distributional analyses of sales taxes look at the final incidence of the tax. This means that our sales tax incidence estimates include not only the sales taxes paid directly by consumers when they buy goods and services at retail, but also the effect of the sales taxes paid initially by businesses. These business sales taxes are, in general, ultimately passed through to consumers.
Any estimate of sales tax incidence that did not include these passed-through business sales taxes would substantially understate the real impact of sales taxes on low- and middle-income consumers. This is because taxes paid by businesses range from a quarter to as much as half of all state and local sales taxes collected each year.
ITEP’s microsimulation model is able to forecast not only the amount of potential taxable spending on a given item by individual consumers, but also the amount of spending by businesses on the same item. The model breaks out this business spending into an amount made by “domestic market” businesses—companies that make most or all of their sales within the state—and “national market” businesses—companies that sell their wares all over the United States. Our incidence analyses of the sales tax typically assign most “domestic market” business sales taxes to in-state consumers, and assign most “national market” business sales taxes to residents of other states.
These passed-through business sales taxes have the effect of increasing the prices paid by consumers for all goods and services, including those subject to sales taxes and those that are exempt. The inclusion of these passed-through taxes is part of the reason why ITEP’s estimates of sales tax incidence can appear to imply that low-income consumers are spending a very large share of their income on taxable items.
In addition to residents and businesses, our incidence analyses also estimate the amount of potentially taxable spending by individual visitors from other states. This means that our analyses of sales tax incidence includes three broad components: resident taxes, which fall on in-state residents; business taxes, which partially fall on state residents and partially are “exported” to consumers in other states; and visitor sales taxes, which are entirely exported to consumers in other states.
Why don’t ITEP’s distributional analyses include a dynamic component? Why don’t ITEP’s analyses of (for example) income tax repeal include the dynamic feedback effect that tax cuts may have on a state’s economy?
There is no agreement among economists on the size, or even the direction, of the economic impact of income tax cuts on a state. This is because the impact depends critically on how these tax cuts are paid for. A tax cut paid for with cuts in state and local government spending could hurt a state’s economy, on balance, if the spending cuts are in areas that endanger the state’s long-term ability to maintain the infrastructure that businesses and individuals rely on. A tax cut paid for with increases in other taxes faces similar problems.
Put another way, the assertion that tax cuts will have a dynamic feedback effect on state economic growth generally only makes sense if you also assert that these tax cuts will not have to be paid for.
There are specific instances in which estimating a dynamic effect is eminently sensible. For example, proposals to increase a state’s cigarette tax by $1.00 a pack are typically introduced not just to raise revenues, but to discourage consumption as well. There is a long-term trend of declining smoking rates among the American population, and it is well documented that especially high cigarette tax rates are effective in discouraging the consumption of taxed cigarettes. For this reason, our incidence analyses of large cigarette tax increases typically factor in the likely reduction in consumption that would be due to the tax hike.
How does ITEP estimate the incidence of corporate income taxes?
It is generally agreed that corporate income taxes, at both the state and federal level, fall primarily on owners of capital. In accordance with this theory, ITEP’s incidence analyses of state corporate income taxes typically distribute the incidence of the tax according to nationwide ownership of capital assets such as stocks and bonds. Because the big and highly profitable corporations that pay the vast majority of state corporate income taxes are multi-state or even multi-national corporations, whose shareholders live around the world, our distributional analysis of a given state’s corporate income tax will assign much of the tax to residents of other states. This is why our incidence analyses show a fairly small impact of a given state’s corporate income tax on residents of that state.
The incidence of the tax in ITEP’s analyses is generally quite progressive, because the vast majority of capital income nationwide is held by the very best-off Americans.
Does the ITEP model have information on residents of each state, or is it a national model?
The ITEP model is built on a very large sample of 750,000 actual federal tax returns. This database of tax returns excludes all personal identification information, but otherwise gives a complete picture of income and demographic information for everyone in the sample. This sample is designed to provide a statistically valid picture of the taxpaying behavior of residents of all fifty states and the District of Columbia. Each taxpayer record in the sample is assigned a weight, representing a specific number of real-world taxpayers. In each state, the sum of the weights is the actual number of residents in each state. This weighted sample is calibrated to provide an accurate estimate of the total income, property, and sales taxes paid by each state’s residents, and each income group with the state, in any given year. ITEP’s model now effectively uses 2019 as its base year.
Why do ITEP’s distributional analyses typically break down the top 20 percent of the income distribution into subgroups, focusing narrowly on the best-off 1 percent, while presenting the rest of the income distribution in groups of 20 percent?
The best-off twenty percent of Americans enjoyed more than half of nationwide personal income in 2010, according to ITEP’s estimates. The best off 1 percent of taxpayers alone enjoyed 18 percent of nationwide personal income. (By contrast, the poorest 20 percent of Americans earned about 3 percent of nationwide income.) This means that incremental differences in the tax treatment of the best-off taxpayers can have substantial implications for state tax collections. Moreover, many states have rules in place that provide special tax breaks for capital gains and other income sources that are highly concentrated in the hands of the best-off 1 percent. An analysis showing the impact of a capital gains tax break on families in the top 20 percent of the income distribution would gloss over the substantial differences in how such a tax break treats the “poorest” and richest members of the top 20 percent.
ITEP’s distributional analyses organize taxpayers by income level. But what definition of “income” is used?
There are two broad ways in which a distributional analysis can sort taxpayers by income level. One approach, used by legislative fiscal analysts in most states, uses income definitions based on “Adjusted Gross Income.” In this approach, the starting point is the income that is actually subject to income taxes in a given state. The other approach, used by ITEP, is to use a more universal income definition, including both income that is subject to tax and income that is exempt.
For components of income that are subject to income taxes, ITEP relies on information from the Internal Revenue Service’s “Statistics of Income” publication, which provides detailed state-specific information on components of income at different income levels. For components of income that are either fully or partially tax-exempt, ITEP uses data from the Congressional Budget Office and the Current Population Survey to estimate income levels in each state. The generally non-taxable income items for which ITEP makes state-by-state estimates (which are included in our measure of “total income”) include: Social Security benefits, Worker’s Compensation benefits, unemployment compensation, VA benefits, child support, financial assistance, public assistance, and SSI.
It’s widely understood that taxpayers at all income levels tend to under-report certain income categories, especially capital gains, pass-through business income, rental income and farm income. For this reason, ITEP’s model estimates the amount of unreported income of each type in each state. These estimates are based on data from a series of Internal Revenue Service research reports that measure the income “tax gap”– the difference between the amounts of various types of income that are reported for tax purposes each year and the (larger) amounts that are actually earned. The tax gap is primarily the result of underreporting by taxpayers. This unreported income is included in our “total income” estimates for each state.
There are two straightforward reasons why our comprehensive approach to measuring income is more informative than the “Adjusted Gross Income” approach. One is that different states use very different income tax rules, so “Adjusted Gross Income” is a concept that means very different things in different states. Another is large amounts of income are excluded from Adjusted Gross Income entirely, even though these income sources represent a meaningful element in a taxpayer’s “ability to pay.” If the goal is to express the impact of a tax change in relation to a taxpayer’s ability to pay it, the only accurate way of doing so is to express these tax changes as a share of total income, rather than Adjusted Gross Income.
How are ITEP’s income definitions aged forward to current-year (or future) levels?
Each year, the Internal Revenue Service publishes state-by-state data showing the levels of federal Adjusted Gross Income (and its components) by income level. ITEP uses these data to “age” our income estimates in each state. ITEP also uses the latest economic forecasts from the Congressional Budget Office to estimate state-level income growth for years in which the IRS data are not yet available. However, simply “aging” the data cannot adequately account for differences in income growth rates between lower- and higher- income taxpayers, and also can’t account for over-time differences in the frequency with which taxpayers report certain types of income such as capital gains or unemployment benefits. For this reason, ITEP regularly adjusts the weights associated with taxpayers at different income levels in each state in a way that amounts to creating a new, updated version of the data. Because this weight-adjustment process has been completed in a way that matches the tax year 2019 data released by the IRS, ITEP’s model now effectively uses 2019 as its base year.
ITEP’s distributional analyses only focus on the impact of taxes, and exclude the distributional impact of state and local spending. Since state governments provide some direct spending targeted to low-income families, shouldn’t these spending-side provisions be included in ITEP’s estimates of the impact of tax systems on low-income families?
State and local governments spend more than $3 trillion a year on various public investments. While some of this spending is targeted to specific income groups in a way that could, with some difficulty, be quantified, much of it is spent in a way that is much more difficult to attribute to specific income groups. For example, close to a third of state and local spending is on education, which directly benefits families with children but also benefits employers who hire products of a state’s educational system. State and local spending on police and court systems safeguards private property, including business, homes and intellectual property.
As it happens, the areas of spending for which it would be least difficult to estimate distributional consequences are the income-security programs that are targeted to low-income families. But states spend far more on untargeted programs that likely are most valuable for the best-off taxpayers.
How do ITEP’s distributional analyses differ from those published by other organizations?
Only three states—Maine, Minnesota and Texas—have the capacity to conduct regular microsimulation analyses of the incidence of their own states. In the rest of the states, state and local lawmakers have no tools at their disposal to analyze these questions, and no other non-governmental organization has this capacity.