Income Tax Compliance: A Unique Experiment in Minnesota
Coleman, Stephen, Government Finance Review
A controlled experiment yielded a rough but realistic estimate of the portion of the tax gap that is truly recoverable through increased voluntary compliance.
Editor's note: The Federation of Tax Administrators (FTA) presented its 1996 Award for Research and Tax Analysis to the Minnesota Department of Revenue in recognition of the department's income tax compliance project. The award is made to revenue departments for research programs demonstrating excellence and importance to tax administration. This article describes the methodology Minnesota used for conducting research on strategies to improve tax compliance and summarizes findings from the research.
In 1995, the Minnesota Department of Revenue conducted an experiment to measure the effectiveness of four strategies to increase voluntary compliance with the individual income tax: 1) an increased examination and audit rate of tax returns with prior notice to taxpayers, 2) enhanced customer service, 3) letters to taxpayers with messages on the importance of voluntary compliance, and 4) form redesign. The experiment measured the impact of these strategies on voluntary compliance by looking at changes from 1993 to 1994 in reported income and taxes paid by groups of taxpayers subject to the different strategies. These changes were compared with changes in a control group of taxpayers who were not affected by the experiment. Approximately 47,000 randomly selected taxpayers were participants in some phase of the experiment; a number of others participated as control groups.
This was a unique experiment; none like it had been conducted by another state or by the U.S. Internal Revenue Service (IRS). To help design the experiment and interpret the results, the Minnesota Department of Revenue set up a five-member advisory board, which included nationally recognized experts in tax compliance research and statistical methods from Minnesota and other states. Representatives from St. Paul's IRS office also participated. Using a special statistical approach, the experiment identifies groups of taxpayers where the compliance strategies have the greatest impact. Results and findings from the experiment provide valuable information about the most effective strategies to increase compliance.
Notice of Examination or Audit
This part of the experiment was designed to test what happens when taxpayers know that their return will be closely examined or audited. The audit or examination threat was intended for two groups of taxpayers: a group representative of the general population of taxpayers, called low risk, and a group believed to be at a higher risk of tax evasion. The sample was stratified so that it included a greater proportion of taxpayers in the high-risk category than one would find in the taxpayer population.
Previous research on tax evasion points to several factors associated with evasion, including income not subject to withholding tax and ownership of a sole proprietorship. The high-risk group was a random sample from taxpayers who filed a federal Schedule C or F (indicating business or farm income) in 1993 and paid Minnesota estimated tax. The estimated tax is another indicator of income not subject to withholding tax. A Minnesota taxpayer is required to file and pay estimated tax quarterly if expected income will be $500 or more above withholding and expected tax credits. The $500 threshold effectively eliminated taxpayers from the high-risk group who may have filed a Schedule C or F but had little income from their businesses.
The advantage of a sample based on payers of estimated tax is the possibility of tailoring interventions for this group in the future if the experiment proved a success, because these taxpayers are involved with the department throughout the year. The low-risk group selected to represent the general population would provide valuable information about whether this approach to compliance works best with people who rarely would be the target of a state tax audit. …