April 5, 2018
When the IRS Visits: Tax Enforcement Methods Matter
U-M researchers find in-person approach leads to higher collections from at-risk companies.
Tax enforcement actions by the IRS can have an impact on the behavior of those directly affected and also some taxpayers connected to them via a network.
University of Michigan tax researchers worked with the IRS to run random experiments that examined effects of letters and in-person visits on how much tax withheld from paychecks at at-risk firms was turned in to the IRS. Depending on the approach, the IRS was able to collect a varying amount of taxes owed, said Will Boning, a doctoral student in economics.
Boning and his colleagues, including Michigan Ross Professor Joel Slemrod, found that on average, the IRS would collect $111,324 when it visited a firm and $13,196 when it sent a letter to collect on withheld income and payroll tax remittances.
"Overall, the noncompliance rate in the population is very low," Boning said. "We are looking at a subgroup of firms that are in trouble."
The U-M researchers studied the effects of visits and letters on both the firms targeted and the ripple effect through their networks of connections through tax preparers and subsidiaries.
The large-scale, randomized field experiment enabled them to examine both the direct and network effects of letters and in-person visits. Visited firms remit substantially more tax.
Because of the random nature of the intervention, the additional remitted tax signals additional tax compliance. Their tax preparers' other clients also remit slightly more tax than usual, while their subsidiaries remit slightly less than usual. Letters have a much smaller direct effect and no network effects, yet may improve compliance at lower cost.
Boning said the research will help the IRS better understand taxpayer behavior and change policy to maximize revenue collection.
"The fact that the IRS is out there actively going after people, that threat of an audit or being contacted about employee tax deposits is very real for people," he said. "Information gets from them to their neighbors. When someone you know gets audited, the possibility that you will goes up in your head."
In addition to Slemrod, Boning's co-researchers included U-M economist Ugo Troiano, and John Guyton and Ronald Hodge of the IRS.