Each year, Her Majesty's Revenue and Customs (HMRC) loses millions of pounds to undeclared income. The department has run several large field studies investigating the effect of Upfront Honesty Prompts (UHDs) on taxpayer dishonesty.
Two studies, on a total of 6,816 people, tested the effect of prompt types on cheating behavior. Participants completed coin flip, die roll, and matrix solving tasks, self-reporting the results of each. When reporting their results, participants could earn extra raffle entries by inflating their scores.
Before a task, randomized participants agreed to an honesty statement by checking a box, signing their name, typing their name, or re-typing the statement: “I commit to providing honest and accurate information in the following task.” The typed name and statement conditions resulted in fewer cheaters than the control. These differences were statistically significant in both studies.
Figure 1. Percentage of dishonest participants by prompt type.
The second study added prompt frequency and comprised three tasks a day for three days. The number of days and times per day that participants received an honesty prompt was randomized, for a maximum of nine total prompts.
Prompt frequency impacted honesty more than prompt type, with cheating decreasing as the number of prompts increased. The multi-task design also competed prompt repetition against prompt recency, finding that repetition had an equal or greater influence on honesty in five of six tests. Practitioners should test moving the most important section of the tax form (in terms of honesty) to the end to increase the number of prompts beforehand.
Figure 2. Percentage of dishonest participants by prompt frequency.
Despite limitations of laboratory studies, the findings on prompt frequency and prompt type have clear implications for policy that deserve further examination:
- Adopt Typed Name – Typed Name should be adopted as the prompt type.
- Adopt Multiple Prompts – Honesty prompts should be added to all areas in tax forms where citizens could lie. Multiple prompts do not have negative side effects.
- Move Important Sections to the End – Tax sections with the highest monetary value (or most expected cheating) should be positioned in places that follow a greater number of honesty prompts.
Test in the Field – Further field experimentation should be employed when implementing these recommendations.