Income Tax Payers Are Not All the Same – A Behavioral Letter Experiment in Eswatini

Very little is known about why taxpayers in sub-Saharan Africa pay their taxes. This article reports results from a nationwide randomized controlled trial in Eswatini, nudging more than 20,000 income tax payers with behaviorally-informed mailings, building on deterrence, facilitation, and trust paradigms. This study is the first to target three different categories of taxpayers at the same time – non-filers, nil-filers and active filers, and targets both companies and individual taxpayers. Most of the literature focuses on active filers. The results show that nudging is very effective with non-filers, especially when controlling for actual collection of the letter – any mailing increases the probability of filing by 1.7 percentage points (p.p.), or 20 percent of the control group mean. Deterrence is particularly effective for non-filing companies – increasing filing by 3.9 p.p., whereas individuals react more to an instructional nudge. Conversely, nil-filers do not respond to a nudge. A trust-based mailing had the opposite of the intended effect with active taxpayers, but they are less likely to nil-file when nudged.