Analysis of Nonresponse in the Statistics of Income's 1999 Individual Tax Return Panel

• Activity Number: 305
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Tara R. Wells*+
• Companies: Statistics of Income/IRS
• Keywords: Nonresponse bias ; Propensity score adjustment ; Panel attrition

Abstract

In 1999, the Statistics of Income (SOI) Division of the Internal Revenue Service began collecting individual tax returns for SOI's 1999 Individual Tax Return Panel. Longitudinal tax data is essential to study how the tax system affects taxpayers over an extended period of time. However, as in all panels, SOI's 1999 Individual Tax Return Panel is impaired by panel attrition. Previous papers have evaluated the presence and motivation for attrition in prior SOI individual tax return panels, yet there has not been research that has measured the nonresponse error caused by the attrition. In this research, I use an exploratory approach to estimate the nonresponse error in specific tax-related variables collected from SOI's 1999 Individual Tax Return Panel from 2000 to 2003. Then I use a propensity score method of subclassification to investigate if the nonresponse bias can be removed. My results show that the nonresponse bias in the tax-related variables can be reduced with the use of propensity score adjustments.