The role of interactions between taxpayers and tax officials in shaping compliance strategies remains unexplored in the literature on tax administration and compliance in developing countries. Nevertheless, practically, taxpayers' experience is shaped by interactions with those who implement tax laws: tax officials. Drawing on a survey of small and medium Rwandan businesses, we provide descriptive and causal evidence into taxpayer experiences and the impact of tax officials’ attitudes and behaviours on taxpayers’ perceptions. A vignette experiment proves that interactions shape compliance attitudes. Facilitation-based interactions improve trust and perceptions of professionalism, while enforcement-focused interactions deteriorate tax morale and respect for the tax administration. Moreover, a population-wide survey of Rwandan tax officials explores how taxpayer characteristics influence officials’ attitudes. A conjoint experiment confirms that tax officials favour more knowledgeable businesses and distrust larger, wealthy taxpayers. The analysis highlights taxpayer knowledge and facilitation efforts as key to improving compliance through healthier taxpayer-administration interactions.

We report the results of a text message campaign to promote tax compliance among landowners in Dar es Salaam, Tanzania. Landowners were randomly assigned to one of four groups designed to test different aspects of tax morale. They received a simple text message reminder to pay their tax (a test of salience), a message highlighting the connection between taxes and public services (reciprocity), a message communicating that people who did not pay were not contributing to local or national development (social pressure), or no message (control).

Recipients of any message were 18 percent more likely to pay any property tax by the end of the study period. Total payment amounts were highest for recipients of reciprocity messages. The average estimated benefit-cost ratio across treatments is 36:1 due to the low cost of the intervention, with higher cost-effectiveness for reciprocity messages. Significant geographic heterogeneity in treatment effect sizes and estimated cost-effectiveness were observed.

The compliance effect of tax agents in low-income countries has received little attention in the literature. This study asses their impact through matching analysis of the universe of CIT and VAT returns submitted in Uganda between 2019 and 2023. Tax agents’ impact, proxied by the presence of audit expenses, is confirmed as broadly positive. CIT returns prepared by agents show no difference in declared liabilities in the aggregate, and higher declared CIT liabilities in the case of small and medium taxpayers. Significant but small increases in total VAT declared are mediated by increases in reported input and output VAT. Taxpayers relying on agents’ services are also less likely to nil-file and more likely to file late, while a higher likelihood of audit selection does not lead to significant differences in audit adjustments. These results are robust to different specifications and an alternative definition of agents’ use based on survey data.

In a natural field experiment with randomised controlled trials, tax debtors received three emails from the Kenya Revenue Authority regarding a tax amnesty. The subjects were randomly assigned to five groups one of which was not contacted. Results show that sequencing nudges in subsequent reminders is effective for tax amnesty uptake. Relative to no email control group, we find average effect size of 9 percentage points on uptake. In addition, sequencing of social norms and deterrence reminders lead to a 2.5 percentage points higher uptake regardless of the treatment order. Deterrence nudges are effective for individuals while payment outcomes are significantly high for early takers. Those who take up the amnesty in the first two rounds pay 30 percent more than late takers. This study extends literature on letter studies by focusing on the context of a tax amnesty and introducing sequenced nudges in subsequent reminders.