Ghostbusting in Detroit: Evidence on Nonfilers from a Controlled Field Experiment

Abstract

Many people who owe income tax fail to file a timely tax return. In communication with these ghosts, what messages from the tax authority are effective for eliciting a return? This is the first study to address message content in communication with income tax nonfilers. I assess the efficacy of messages related to penalty salience, punishment probability, compliance cost, and civic pride by evaluating the response to experimental mailings distributed by Detroit to 7,142 suspected resident nonfilers. The penalty salience message was the most effective. Relative to a basic mailing that requested a return, penalty salience mailings that stated the statutory penalty for failing to file a return tripled response rates from 3% to 10%, increased the number of back-year returns filed per response from 0.08 to 0.27, and raised the fraction of filed returns that admitted tax due from 39% to 52%. Compliance cost mailings that enclosed a blank tax return and punishment probability mailings that stated the recipient's federal income also raised response rates relative to the basic mailing, but civic pride mailings did not. Mailings were more effective in eliciting returns from older, higher-income, and first-time nonfilers. I investigate the impact of treatment mailings on the behavior of untreated neighbors and find no evidence of geographic network effects.