Saturday, November 9, 2013
3016 Adams (Washington Marriott)
*Names in bold indicate Presenter
This paper examines the payment of income and payroll taxes from a longitudinal perspective. Several studies have used cross-sectional tax return data to estimate the fraction of families that on net pay into (or receive benefits from) the income and payroll tax systems in the United States at a point in time. While informative, such studies suffer from several weaknesses. First, a taxpayer may be a beneficiary in one year but a net payer the next (and vice versa), making it unclear whether such a division between payers and beneficiaries is similar over longer time horizons. Second, using families as the unit of observation does not account for the fact that payer or beneficiary status may be correlated with family size, and so the fraction of families in beneficiary or payer status may differ from the fraction of individuals. In this paper, we construct a panel of tax information at the individual level that allows us to address the question of who pays taxes over longer time frames. To create this panel, we compiled a random sample of tax returns over the 2001-2011 period, and matched and merged these returns to a panel of IRS information returns (including forms W-2, 1098, 1099, and others) that spans the same period. Because the information return dataset includes forms from non-filers, the resulting dataset represents a random sample of all individuals about whom the IRS received any information in any year of our sample. To verify consistency with currently available tabulations, we examine what fraction of families are net payers and beneficiaries on an annual basis using Current Population Survey information to account for the number of non-filer families. Next, using our individual level data, we construct a net payers and beneficiaries measure (on an annual basis) that does not require any imputation for the non-filers, and probe whether this division differs by age cohort. We then utilize the panel structure of our data to examine how many years in a given period an individual is a net payer or beneficiary. Further, we examine the likelihood that an individual switches to being a net payer in subsequent years, conditional on being a net beneficiary in a given year. Finally, we examine the extent to which the total amount paid over a given period exceeds the net benefits received over that same period.