Methodology for Pew’s Tax Expenditure Database
Subsidyscope presents estimates of annual income tax expenditures by the Department of the Treasury in the President's Budget, and from the Congress' Joint Committee on Taxation (JCT) in its report "Estimates of Federal Tax Expenditures." Each estimate indicates the amount of revenue that the government loses (not taking into account behavioral responses that might occur when eliminating one tax expenditure leads taxpayers to take another) due to specific provisions in the tax code, such as exemptions, credits and deductions, exclusions, preferential rates or deferrals. The estimates presented are derived using tax models.
Summing Tax Expenditure Data
Each tax expenditure is estimated independently of other tax expenditures and behavioral responses, meaning interaction effects are not taken into account. The Treasury and JCT differ in the way they treat interactions among tax expenditures. JCT assumes when calculating each tax expenditure that if one is eliminated, the taxable entities involved will fall back to their next best tax option, which may include using other tax expenditures to reduce their liability. In contrast, the Treasury does not assume that eliminating a tax expenditure would prompt a taxable entity to make more use of other tax expenditures.
Therefore, when aggregating groups of estimates together, or summing total tax expenditures, the sum may be different, depending on whether interaction effects are included. As such, the aggregations presented should be viewed as approximations and not the exact amount of revenue that the government would bring in if any of the tax expenditures were eliminated. (For more on summing tax expenditures and interaction effects, see Burman, Toder and Geissler, 2008).1
Presenting Data in Pew's Tax Expenditure Database
The Subsidyscope data has been compiled from the Treasury and JCT tax expenditure budgets. The report years, presented along the vertical axis (down the rows), correspond to the year of the budget or document in which the estimates were obtained. The estimate years, presented along the horizontal axis (across the columns), correspond to the years of estimates within a specific document. Subsidyscope has not adjusted these estimates for inflation.
Matching Treasury and JCT Tax Expenditures
Subsidyscope presents the estimates together when their definitions and budget functions match.2 In some cases where they do not, Subsidyscope researches the content of the estimates and, where appropriate, matches the estimates. In other cases, even after researching the tax expenditures, it is unclear whether a Treasury and JCT estimate are describing the same one. In these cases, the tax expenditures are left unmatched. The definitions used to match data can be found in the "Tax Expenditures" chapter of the "Analytical Perspectives," found in each of the President’s Budgets (for the Treasury tax expenditures), and in the Congressional Research Service's "2010 Tax Expenditures: Compendium of Background Material on Individual Provisions" (for JCT tax expenditures).3
For presentation purposes, Subsidyscope generally uses Treasury's naming convention to describe tax expenditures, and when a name changes, Subsidyscope uses the most recent. Once these are matched, the JCT tax expenditure is listed under the most recent Treasury title. When the Treasury and JCT tax expenditure line items match, but are found in different budget functions, a note is added to the descriptions of both line items to indicate where the corresponding tax expenditure can be found. Sometimes, one of the agencies will use a broader definition to estimate certain tax expenditures than the other; for example, JCT might break out specific provisions of a tax expenditure and estimate them separately, whereas Treasury will estimate them as a group, and provide only one total. When this occurs, and it is determined that the tax expenditures are the same, Subsidyscope presents the entire group of Treasury or JCT tax expenditures next to the corresponding Treasury or JCT tax expenditure. When there are multiple Treasury tax expenditures matching one JCT tax expenditure, the most recent JCT naming convention is used.
Accuracy of the Data
Subsidyscope has made every effort to ensure that the information in its database replicates that in the corresponding Treasury and JCT budgets. Because Subsidyscope relies on government data, any errors or omissions in those sources will be replicated in the Pew data set. In rare cases, described below, the database information might differ slightly from the source. Wherever possible, Subsidyscope has noted these differences where they occur. Both the Treasury and JCT provide separate estimates for individuals and corporations, and Subsidyscope uses these to populate its database. Although the Treasury also provides tables with the individual and corporate estimates combined, the data in Subsidyscope's database are created by summing the separate individual and corporate data (rather than importing the data already combined by the Treasury). There are some instances in which the combined total in the Treasury report does not equal the sum of the estimates. In these instances, the Subsidyscope number will differ from the Treasury document, and Subsidyscope notes this in the database.
For both the Treasury and JCT, Subsidyscope presents footnotes as they are shown in the government documents. In some cases in which footnotes appear to be missing or incorrect, Pew includes the original footnote and a suggested correction.
The JCT does not provide specific numerical values for positive tax expenditures of less than $50 million, or negative tax expenditures of greater than negative $50 million. Instead, it generally notes that a positive or negative tax expenditure of less than $50 million or greater than negative $50 million is being estimated. In its database, Subsidyscope presents these values using an asterisk. The asterisk does not distinguish between a positive tax expenditure of less than $50 million or a negative tax expenditure of greater than negative $50 million, although this detail is provided in the downloadable comma-separated values (.csv) file available on each tax expenditure's unique page. When multiple positive or negative tax expenditures of less than $50 million or greater than negative $50 million are aggregated, the total value is unknown. Thus, the asterisk is presented next to aggregated values in the database in these cases.
Comparing the Database and the Downloadable CSVs
Each page of Pew's Tax Expenditure Database can be downloaded into a .csv file through a link in the top right-hand corner of the page. The data presented replicates the data on the particular Web page, with one exception. The JCT values of less than $50 million or greater than negative $50 million, which are represented as an asterisk in the database, are presented slightly differently in the downloadable .csvs, as elaborated below:
- The .csvs do not contain asterisks. In .csvs with individual tax expenditure data, the estimates in the separate individual and corporation columns that contain values between negative $50 million and $50 million are denoted as either ">-50" or "<50" (whereas in the database, they are denoted by an asterisk). These are the only .csvs in which you will find these values presented. When aggregated in the totals columns, the sums of these values are unknown and thus are rounded to zero.
- In all other .csvs, such as those found on the sector pages, which include the individual tax expenditures contained in each sector, the JCT values between negative $50 million and $50 million are rounded to zero. When aggregated in the totals columns, the sums of these values are also unknown and thus are rounded to zero. To determine if an estimate's value is "greater than negative $50 million" or "less than $50 million" please download the .csvs with individual tax expenditure data.
In addition to the downloadable CSVs on each database page, Subsidyscope has created downloadable .csvs that contain the Treasury and JCT data in its tax expenditure database, broken out by budget year. To access these, see the Datasets page.
Differences between the Treasury and JCT Tax Expenditure Estimates
By presenting the Treasury and JCT estimates side-by-side, Subsidyscope's database enables quick comparisons of the estimates produced by each agency. Database users should be aware of several reasons why Treasury and JCT estimates might differ. Sometimes there are presentational differences that make the estimates appear more divergent than they actually are. For example, see Table 1 here.
There are also methodological differences in estimates produced by the Treasury and JCT, which arise due to baseline structures used to define tax expenditures. For instance, one difference between the Treasury's and JCT's approaches is in the baseline against which they measure tax expenditures. Since 1983, the Treasury primarily has used a "reference baseline" that reflects the major provisions of the tax code.4 JCT, following earlier precedent and the provisions of the Congressional Budget and Impoundment Control Act of 1974, uses in its baseline a notion of "normal income tax base," which is a somewhat broader definition of potentially taxable income than current law uses. (JCT has critically examined the judgments that underlie the concept of a "normal income tax base" in a methodology paper). In its tax expenditure budget, JCT lays out the differences between the Treasury and JCT methodologies, which Subsidyscope presents below.
- Burman, Leonard, Eric Toder and Christopher Geissler. "How Big Are Total Individual Income Tax Expenditures, and Who Benefits from Them?" The Urban Institute. Washington, DC. December 2008.
- The Treasury and JCT group tax expenditures together using budget functions. The federal budget is divided into approximately 20 categories called budget functions. These categories include all spending for a given topic, regardless of the federal agency that oversees the individual federal program. Subsidyscope presents tax expenditures according to the budget function into which the federal government has placed them.
- When a tax expenditure cannot be found in "2010 Tax Expenditures: Compendium of Background Material on Individual Provisions," definitions are taken from "2008 Tax Expenditures: Compendium of Background Material on Individual Provisions," and "2006 Tax Expenditures: Compendium of Background Material on Individual Provisions."
- Joint Committee on Taxation. "A Reconsideration of Tax Expenditure Analysis." JCX-37-08. Pg. 25.
- If the exclusion were repealed, the value of the employer-paid health insurance would be included in income and taxpayers would be treated as having purchased the insurance themselves. Thus, the insurance expense would be deductible as an itemized medical expense on Schedule A, subject to the itemized medical deduction floor (7.5 percent of the taxpayer's adjusted gross income).