The ITEP Tax Microsimulation Model is used by ITEP to produce federal, state, and local distributional analyses and revenue estimates. The ITEP Model was originally developed in 1996 and has consistently undergone updates and enhancements to incorporate newer data and take advantage of more advanced computing technology.
The model builds on the Internal Revenue Service (IRS)’s most recent Public Use File (PUF), enhanced by the Census Bureau’s American Community Survey, the Consumer Expenditure Survey and numerous other data sources. Data are extrapolated to later years using detailed IRS state-level data as well as other official economic data and projections.
The model computes the revenue yield and incidence of both current tax law and proposed tax law changes. It is unique in its ability to produce analyses at the federal and state levels that comprehensively examine income, consumption, and property taxes. It is also the only tax microsimulation model capable of disaggregating tax incidence data by race and ethnicity at the state level.
The PUF, the starting point for the model, is an anonymized database that contains substantial detail on over 200,000 tax returns representative of the full federal filing population. In addition to the high-level variables that appear on Form 1040, the PUF provides a wealth of information from other forms covering itemized deductions, capital gains components, sources of private business income, determinants of credit eligibility, and other related taxes such as the Alternative Minimum Tax (AMT) and the Net Investment Income Tax (NIIT). The IRS makes PUF files available with a delay to allow the agency time to ensure that tax return confidentiality is maintained. The ITEP model uses the 2015 PUF, the latest year available.
To allow for state-level analysis, the federal records from the PUF are assigned to states using a Poisson regression-based weighting model that ensures that each state’s filing population matches published IRS state-level data by income band across 63 variables covering filing status, dependents, sources of income, adjustments, deductions, credits, and tax liability. Weights assigned in the model indicate the number of real-world tax units in each state represented by each record. The model also contains local-level weights for certain local jurisdictions of interest.
Tax return data from the PUF are paired with observations from the U.S. Census Bureau’s American Community Survey (ACS) to create a valid representation of the U.S. population, including federal filers and non-filers, both nationally and for every state. A statistical match with the ACS allows ITEP to gain access to data on demographics and non-taxable income sources not available on tax returns. The entire dataset is also supplemented with imputed values based on econometric analysis of other datasets such as the Bureau of Labor Statistics’ Consumer Expenditure Survey and the Federal Reserve’s Survey of Consumer Finances.
Each record in the model includes components of personal income and of consumer spending, the value of real and personal property, and a wide range of other tax items and demographic and social characteristics. The weights and dollar amounts on each record are aged and extrapolated forward using a linear programming routine to allow the model to match state-by-state and national targets by income band across 93 variables taken from the most recently published state-level IRS tables. To bring the data up to the current tax year and those beyond, model values and weights associated with each record are aged to reflect the most recent data and projections published by the IRS, the Congressional Budget Office, the Census Bureau, the Bureau of Economic Analysis and state revenue offices, with variation across states to reflect differences in the trajectories of population and economic growth.
The ITEP Tax Microsimulation Model is divided into a set of modules, each of which focuses on a particular type of tax. Each module consists of two components. The first is the microdata set described above. The second is a set of calculators reflecting existing and proposed laws applied to each record in the microdata.
The Personal Income Tax Module analyzes the revenue and incidence of current federal, state, and local personal income taxes and policy options. The ITEP Model’s personal income tax module contains a set of tax calculators reflecting both past and present federal, state, and local personal income tax laws. These calculators were developed by ITEP staff through careful review of tax forms, statutes, and recently enacted legislation. Applying those calculators to the detailed microdata described above yields distributional results and tax revenue estimates for the jurisdiction being analyzed.
The Consumption Tax Module analyzes the revenue yield and incidence of taxes levied on the purchase of goods and services and on business gross receipts. There are more than 700 tax base items available in the item database, which allow for accurate modeling of the wide array of different sales tax bases used at the state and local levels, as well as various smaller taxes levied on narrow categories of spending. The consumption module also includes supplementary data on visitor consumption, reflecting the in-state purchases of individuals residing in other states or countries, and for business consumption, reflecting the in-state purchases of both capital goods and short-lived intermediate inputs by businesses competing in local and national markets. The tax calculators in this module reflect the sales tax policies of a range of national, state, and local jurisdictions. Applying those calculators to the database yields distributional results and tax revenue estimates for the jurisdiction being analyzed.
The Property Tax Module analyzes the revenue yield and incidence of current state and local taxes on real and personal property. It can also analyze the revenue and incidence impacts of statewide policy changes in property taxation, including the effect of circuit breakers, homestead exemptions, and other tax reduction devices. Each record in the microdata set includes real and personal property data such as homeowner status, home value, rent, vehicle count, and vehicle value. The tax calculators in this module reflect the property tax policies of a range of state and local jurisdictions.
The structure of the ITEP model’s income tax module mirrors models at the federal level maintained by the congressional Joint Committee on Taxation, the U.S. Treasury Department, and the Congressional Budget Office, and at the state level by the Minnesota Department of Revenue and other state agencies. Microsimulation modeling is widely regarded as the best approach to tax policy analysis because of its ability to account for overlapping and interacting tax provisions and to produce results that are representative of the full population. The ITEP model stands apart in its ability to produce accurate incidence and revenue estimates for every state in the country, and for the integration of the income tax model with property and consumption tax models, allowing a comprehensive assessment of the overall progressivity and revenue capacity of state tax systems. The ITEP model also offers the unique capacity to disaggregate state level incidence analyses by race and ethnicity, and to analyze incidence at the local level for select jurisdictions.
ITEP TAX MICROSIMULATION MODEL DATA SOURCES
The ITEP Model uses the following microdata sets and aggregate data:
IRS Individual Public Use Tax Files; American Community Survey Public Use Microdata Samples; IRS Statistics of Income, Individual Tax, By State; IRS Statistics of Income, Business Tax, National and By State; Bureau of Labor Statistics Consumer Expenditure Survey; Census American Community Survey tabular data; Survey of Income and Program Participation; Panel Study of Income Dynamics; Survey of Consumer Finances; miscellaneous IRS data; state tax, budget, and fiscal agency data from all 50 states and the District of Columbia; state assessors data; Census Government Finance data; Congressional Budget Office and Joint Committee on Taxation forecasts; Bureau of Economic Analysis (BEA) Gross Domestic Product by State and National, by Industry; BEA Personal Consumption Expenditures National and by State; BEA Fixed Assets By Industry; BEA Input-Output Accounts Data; American Housing Survey; Census of Population Housing; Energy Information Administration; state transportation department data; Federal Highway Administration (FHWA) Highway Statistics series; FHWA’s National Household Travel Survey; U.S. Department of Transportation Bureau of Transportation Statistics; National Equity Atlas (produced by PolicyLink and the USC Program for Environmental and Regional Equity); Census County Business Patterns; U.S. Department of Agriculture National Household Food Acquisition and Purchase Survey; U.S. Travel Association reports; American National Election Studies; National League of Cities; Centers for Disease Control and Prevention; Kaiser Family Foundation; Rental Housing Finance Survey; Zillow Price-to-Rent Ratio data; EY reports; Lincoln Institute of Land Policy and Minnesota Center for Fiscal Excellence reports; Current Population Survey Tobacco Use Supplement; Orzechowski and Walker’s Tax Burden on Tobacco; Distilled Spirits Council data; state liquor agency data; American Gaming Association; U.S. Department of Education National Center for Education Statistics; Insurance Information Institute; Society of Actuaries data; Federal Reserve Board, Financial Accounts of the United States.
