Unwilling or Unable to Cheat? Evidence From a Tax Audit Experiment in Denmark

The experiment was conducted in two phases. In 2007, a random sample of 19,600 taxpayers had their tax returns fully audited. In 2008, after filling their tax returns, but while they were still able to change them, taxpayers received a letter from the tax authorities assigning them randomly to one of three groups: certainty of audit, 50% chance of audit, or control group (no letter).

The results show that less than 1% of third party reported income is evaded, while more than 40% of income not subject to third party reporting is evaded. However due to the intensive use of third party reporting by the Danish tax authorities, total evasion is estimated at less than 3% of tax liabilities. Other results show an important subsequent effect of having been audited in 2007 on 2008 report - a positive monotonic increase in reported income following an audit threat. The authors also found that the effect of marginal tax rates on evasion is very low which indicates that rigorous tax enforcement is a better tool at combating evasion than manipulating marginal tax rates.

This article presents one of the first randomised studies on fiscal evasion. Its clever design and large sample allows for an estimation of responses by taxpayers to previous audits and audit threats. Finally, it highlights the importance of third party reporting to minimise tax evasion, rather than national sentiment or pure honesty as suggested by previous literature.

H.J. Kleven et. al, “Unwilling or Unable to Cheat? Evidence from a Tax Audit Experiment in Denmark”,  Econometrica,  79: 651–692, 2011