Rwanda's tax authority found threats backfired on best payers

By the mid-2010s Rwanda's tax authority, like most revenue agencies, relied on the standard tool for chasing under-declared corporate tax: a letter warning of audit risk and penalties if the taxpayer did not correct their filing. Compliance research elsewhere had started to question whether deterrence messaging was actually the most effective lever, but almost none of that evidence came from a low-income country's tax administration, where audit capacity is thin and taxpayers may reasonably doubt any threat will be followed up.

Rather than assume the deterrence letter worked because it was what everyone sent, the Rwanda Revenue Authority partnered with researchers from the International Centre for Tax and Development to run the question as an experiment on real taxpayers rather than a policy debate. In early 2016 it selected 9,000 taxpayers and randomly assigned them to receive one of nine message variants — three contents (deterrence, an explanation of what tax revenue funds, or a plain reminder) crossed with three channels (letter, email, SMS) — or nothing at all.

Tested nine message-and-channel combinations against a control, finding explanations of tax-funded services beat deterrence, which backfired on top payers.

Random assignment to 9,000 real taxpayers produced causal effects on declared revenue, not assumptions.

Explaining tax benefits lifted compliance 12.3% vs control, beating deterrence's 7.2%.

Deterrence backfired among higher-income taxpayers, a subgroup effect only visible with a large, varied sample.

The experiment cost a few thousand dollars but yielded $9 million in extra declared revenue.

Test your default compliance lever against alternatives; the threat may be weaker and even counterproductive for your best segment.

Messages explaining what tax funds increased compliance by roughly 12.3% versus the control group, ahead of deterrence messaging's 7.2%, and deterrence produced a documented backfiring effect among higher-income taxpayers even as it worked as intended for the rest. Across the full set of interventions the RRA reported close to $9 million in additional declared revenue against a few thousand dollars in messaging costs. A companion study found the effect faded within a year, so the RRA treats it as a cheap, repeatable nudge rather than a one-time fix.