INTRODUCTION

This section describes the methodology and procedures which were developed to estimate taxes paid for the March 1984 CPS microdata files. In all, four types of taxes were simulated: 1) Federal individual income taxes, 2) State individual income taxes, 3) property taxes on owner-occupied housing, and 4) payroll taxes.

Development of the after-tax simulation procedures began with the March CPS annual demographic supplement. This microdata file contains demographic and economic information for approximately 59,000 sample households and the persons living in these households. It includes detailed information on household and family relationship; age; marital status; race and ethnicity; educational attainment; weeks and hours worked during the calendar year; occupation, industry, and class of worker of the job held longest during the calendar year; and income amounts for wages and salary, nonfarm and farm self-employment income, interest, dividends, rental income, estates and trusts, royalties, pension income, unemployment compensation, and sources of nontaxable income as described in appendix B.

The second major element in the simulation system were statistical summaries of individual income tax returns compiled by the Internal Revenue Service. These statistics are made available in the IRS publication series, Statistics of Income (SOI). Some unpublished statistical summaries from the IRS were also used to develop these procedures.

A third element was the 1983 Annual Housing Survey microdata file. This element was used to assign property taxes paid to the March CPS sample households residing in owneroccupied housing.

Finally, in order to estimate proportions of tax filers owning homes and itemizing deductions, tabulations were made from Interview No. 5 (6) of the 1979 Income Survey Development Program.

The system for estimating taxes paid and after-tax income created a modified March CPS microdata file. This file was formed by expanding the March CPS format to include variables relevant to the simulation of taxes paid. The detailed tables contained in this report were derived from this modified March CPS data file.

Federal Income Taxes

Simulation of Federal income taxes required up to four separate operations. First was the formation and classification

of tax filing units using household relationship, marital status, and dependency rules. Second, was the calculation of adjusted gross income for each of those units. Third was the simulation of amount of Federal income taxes paid. Finally, the calculation of earned income tax credits was made, when applicable.

Formation and classification of Federal income tax filing units. A Federal tax filing "unit" was defined as any individual (or married couple) with either $400 in self-employment income, $1,000 in wages or salary, or a total of $1,000 in interest, dividends, rents and royalties, estates and trusts, or pension income in 1983. These income levels were chosen because they either corresponded to tax laws or helped bring the estimated number of filing units on the CPS in line with 1983 IRS Statistics of Income (SOI) data.

The next step in the formation of Federal tax filing units was the assignment of dependency status. The algorithm for assigning dependency for each tax unit used the following rules:

• All filing primary family householders and spouses were included as dependents on their own tax returns.

• All children under age 15 who were members of the primary family were counted as dependents on the return of the family householder. Children aged 15 and over (except related subfamily members) with a total taxable income of less than $1,000 were assigned dependency to the tax return of the primary family householder. Children aged 15 and over who were students were assigned dependency to the primary family householder regardless of income level.

• All other primmary family members (except related subfamily members) with taxable income of less than $1,000 were assigned as dependents on the tax return of the primary family householder.

Related subfamilies having at least one Federal tax filing unit were treated separately in the same manner as primary families. Members of a related subfamily containing no Federal tax filing unit were assigned dependency to the tax return of the primary family householder.

• All unrelated subfamilies were treated in the same manner as primary families.

• Primary and secondary unrelated individuals age 15 and over were treated as dependents only on their own tax returns.

All simulated filing units were classified into one of three return types. Married couples and persons whose marital status was ''married, spouse absent in Armed Forces" were assumed to file joint returns. Unmarried family householders with dependents were assumed to file head of household returns. All other persons classified as Federal tax filing units were assumed to file as single individuals.

Computation of adjusted gross income. Adjusted gross income (AGI) for each simulated tax filing unit was calculated by summing the income amounts from all taxable sources and an imputed amount for capital gains. The sources of CPS income included in AGI were wages and salaries, net farm and nonfarm self-employment income, net rental and royalty income, dividends, interest, estates and trusts, and income from private and government pensions.

Capital gains were imputed to tax filing units based on data obtained from a Statistics of Income (SOI) public use file and reports summarizing information reported on Federal tax returns. These data provide estimates of the probability that a filing unit in a given matrix cell reported capital gains and the mean amount of capital gains for that cell. The variables in this probability matrix were: level of AGI, type of return, and age of tax filer. A Monte Carlo technique was used to randomly assign capital gains: a random number (between O and 1) was generated for each filing unit; if that number was less than or equal to the probability of filing units in that matrix cell reporting capital gains, the mean amount of capital gains, as computed above, was added to that unit's AGI. This procedure does not control on demographic and other characteristics which might affect the allocation of this source of income.

In the calculation of adjusted gross income, a portion of unemployment compensation was also included in AGI if the sum of AGI and unemployment compensation for that tax unit exceeded $12,000 ($ 18,000 for joint returns). In these cases, the lesser of 1) the amount of unemployment compensation or 2) one-half of the difference between the sum of AGI and unemployment compensation and the income limit was included in AGI.

In 1983, married-couple filing units in which both spouses had earnings were allowed to deduct 10 percent of the earned income of the lesser-earnings spouse (to a maximum of $3,000). This new adjustment is reflected in the 1983 tax model. In addition, payments to Individual Retirement Accounts (IRA) were simulated for the 1983 tax model. The May 1983 CPS pension supplement was used to estimate probabilities of tax-filing units contributing to IRA's and the average amounts contributed. These probabilities were then used to assign IRA contributions to individual tax-filing units on the CPS file. The IRA payments were deducted from the total income received by the tax-filing units in order to compute ad

Computation of taxable income and taxes paid. Taxable income was computed by subtracting the estimated allowable deductions from AGI. The first step in this process consisted of predicting which filing units itemized deductions. Homeownership was determined to be the most important variable available from the CPS for assigning itemization status to tax filers. Outlined below is a step-by-step description of the procedures used to assign itemization status.

1. A statistical match was made of the March CPS and Annual Housing Survey (AHS) data files in order to assign a monthly mortgage amount and a property tax amount to each owner-occupied unit on the March CPS file.'

2. Probabilities of itemizing for homeowner, tax-filing units were computed by size of monthly mortgage payment from the 1979 Income Survey Development Program (ISDP) test panel. Probabilities for renters were computed by AGI level.

3. The probabilities described in step 2 were used to randomly assign itemization status within monthly mortgage (or AGI) intervals using the same Monte Carlo technique used in the assignment of capital gains.

4. The amount of itemized deductions for tax filing units was computed using a matrix showing the ratio of itemized deductions to AGI for all units by AGI interval, type of tax return, number of dependents, and presence of a home mortgage. The ratios of itemized deductions to AGI were computed using a 1980 SOI public use file and data published in the 1983 SOI report.

Next, a standard deduction was estimated for each tax filing unit by multiplying the number of exemptions by $1,000. Taxable income was then estimated by subtracting the itemized and standard deductions from AGI. Tax liability was then computed using the appropriate tax schedule for that simulated return type.

The dependent child care credit was simulated for the 1983 Federal tax model and subtracted from the total tax liability. This credit allows tax filers to deduct a portion of child care expenses while they work or look for work. Data from the June 1982 CPS supplement were used to estimate probabilities of tax filers paying for child care.

The simulation procedures do not capture variations in proportions of income paid in taxes within AGI intervals. The proportion of income paid in taxes for households with similar AGI amounts may differ relative to factors such as race, age of household members, number of household members, and marital status. The extent to which these variations exist has not been measured, therefore, caution should be used when interpreting relatively small differences between the incomes of various subgroups of the population.

'A detailed description of the CPS-AHS statistical match can be found

Computation of the earned income tax credit. Earned income tax credits were simulated for the 1983 tax model. These tax credits were used in the calculation of net Federal tax liability and computation of after-tax household income for filing units with one or more dependent children, less than $10,000 in AGI, and earnings between $1 and $10,000.

State Individual Income Taxes

There were 44 States that required payment of individual income taxes in 1983. For the purpose of this model, the definitions of tax filing units and AGI used for the estimation of Federal income taxes were also used for the simulation of State income taxes.

The amounts of State individual income taxes paid were computed by developing a model of each State's income tax regulations. Information on the State tax systems was obtained from a publication entitled State Tax Handbook, October 1, 1984. While every detail of each State's income tax system was not simulated, most of the important aspects were accounted for.

Property Taxes on Owner-Occupied Housing

In 1983, property taxes were estimated using a data file created by the statistical match of the March 1984 CPS and the 1983 AHS. In that statistical match, property tax amounts reported on the 1983 AHS for owner-occupied housing units were assigned to CPS households with similar characteristics (as defined by the matching variables). There was no comparable data file from the AHS for 1982. Property taxes in 1982 were estimated in a two-step process. First, the March 1982 and March 1983 CPS files were statistically matched. The March 1982 property tax amounts (those taken from the 1981 AHS) were then assigned to March 1983 CPS households.

Second, these 1981 amounts were increased based on the rate of increase between 1981 and 1982 in the Bureau of Economic Analysis's figures for residential property taxes adjusted to reflect the increase in the number of households. Property taxes paid on secondary residences such as vacation homes, could not be simulated. Also, the proportion of rent that pays the property taxes on renter-occupied housing units was not estimated.

The estimation procedures for property taxes paid by homeowners produces estimates that do not correspond precisely with those available from the AHS. These differences are mainly the result of differing universes and use of the statistical matching procedure. The published AHS estimate for property taxes is based on a universe that excludes condominiums, cooperatives, and mobile homes, the simulated universe includes these cases. In 1981, the published AHS estimate of property taxes was $671, compared with the

Payroll Taxes

The Social Security payroll tax (FICA) and the Federal Employee Retirement tax were simulated using occupation of longest job and earnings data reported on the CPS. Social Security payroll taxes were calculated directly from the reported CPS earnings using the Social Security payroll tax formula for 1983. For wages and salary, the tax rate used was 6.7 percent up to a maximum of $35,700.

The tax rate for self-employment was 9.35 percent of the amount between $400 and $35,700. Not all workers were assigned coverage under Social Security and, therefore, a small number were not subject to Social Security taxes. All Federal employees and specific proportions of workers in certain occupation groups were assigned noncovered status. Unpublished statistics supplied by the Social Security Administration were used to make these assignments.

Retirement taxes paid by each Federal employee were simulated by multiplying their wages and salary amount by the 7.0 percent tax rate. The identification of Federal employees was based on the class of worker of longest job as reported on the survey.2

COMPARISON OF SIMULATION RESULTS WITH DATA FROM IRS AND OTHER INDEPENDENT SOURCES

The procedures described in the preceding section were translated into a computer simulation model. Tables A-1 through A-4 in this section provide a basic evaluation of the accuracy of this model by presenting comparison of the simulation results with data from independent sources.

Number of Federal Tax Filing Units and Amount of Adjusted Gross Income

Shown in tables A-1 through A-3 are comparisons of IRS and CPS distributions of adjusted gross income and number of returns with specified income types. The 1983 CPS tax simulation yielded 96.1 million Federal tax filing units, about the same as the 1983 preliminary IRS Statistics of Income figure of 96.3 million. The CPS simulated aggregate adjusted gross income was $1,950.4 billion, which was not significantly different from the preliminary IRS figure of $1,950.8 billion. While the CPS and IRS adjusted gross income amounts are not significantly different, there are major differences in the components of total adjusted gross income. Although the IRS data indicate a larger amount of wages and salary income and interest income than the CPS, the CPS recorded signifi

2According to the National Income and Product Accounts published by the Bureau of Economic Analysis (BEA), neither Social Security (FICA) nor Federal Employee Retirement payments are treated as taxes. Instead, they are both included under Federal Government receipts as "Contributions for Social Insurance." We have included them under the broad heading of taxes here for convenience as both are mandatory deductions