Appendix B: Federal and State Income Taxes

Because taxes are paid out of income, policies that increase pre-tax, pre-transfer income also should increase tax revenues. State and federal income taxes move directly with income. To estimate the relationship between educational attainment and tax revenues, we first estimate the taxes paid using income data from the March Supplement to the Current Population Survey, 2000. The base for general tax revenues is pre-transfer family income. Internal Revenue Service (IRS) data on the percentage of total taxes paid by each income group are used to estimate the state and federal income taxes paid by each family in the sample. Table B.1 summarizes the average tax rates applied to each income group. Estimated tax payments per family are then divided evenly among all family members to estimate payments per individual. An ordinary least squares regression model is used to estimate the relationship between education and the natural log of state and federal taxes for the adult sample, while controlling for other factors affecting the level of taxes paid. So, while children were included in the division of taxes among all family members, the final estimates are a result of the adult sample only. Median family income for the sample is approximately $46,000. The median yearly per person state and federal taxes paid are approximately $535 and $1,820, respectively.

Table B.1:  Average 1997 Federal and State Income Tax Rates by Income Class

Two ordinary least squares models were used to estimate the relationship between federal and state taxes and the explanatory variables of education, age, gender, and race and ethnicity.(1) The natural log of individual income tax payments were used as the dependent variables for both models. The following models estimate the amounts of federal and state income taxes, given the explanatory variables: 1) ln(Tj) = Xβ + μ μ ~ N(0,σ2μ) where ln(Tj) is the natural log of federal taxes if j=1 and state taxes if j=2, X is a vector of explanatory variables, β is a vector of estimated coefficients, and μ is a random error term distributed normally with mean zero and variance of σ2μ. Table B.2 contains the parameter estimates of the least squares regression analyses used to estimate the independent relationship between income tax payments and each of the explanatory variables used in the models.

Table B.2:  Least Squares Regression Parameter Estimates for Federal and State Taxes

To view a list of all chapters in this book, click here.  To read the chapters in sequential order, please follow the arrows below.

   Back to Appendix A: Kentucky Survey Data and Models

  Ahead to Appendix C: Criminal Justice System

Footnotes

1. For more detail on the explanatory variables used in this section see Appendix A. Return to text.