Research

Fertility is key to women’s economic and social standing in Africa, and the fear that medical interventions will cause infertility is pervasive. This paper links this fear to the low adoption of hormonal contraceptives—a health technology that provides women with greater control over birth timing and facilitated greater human capital investment and female labor force participation in the West. Focusing on Zambian women pursuing tertiary education, who have a reported desire to delay pregnancy but low contraceptive usage, we implement a novel intervention to change these beliefs. This intervention persistently increases contraceptive use over 6 months and reduces reported pregnancies by as much as 71%. IV estimates indicate that eliminating the belief that contraceptives cause infertility would triple contraceptive use, more than halving the unmet need for contraceptives. Policies that account for the cultural context can successfully increase the use of modern health technologies in hard-to-reach populations.

Can knowledge of buyer-seller networks be leveraged to improve tax compliance? We explore this question through a randomized controlled trial in Uganda. Using transaction-level VAT data, we construct firm networks and identify discrepancies in amounts reported by trading partners. Enforcement letters highlighting these discrepancies are sent to either the seller, the buyer, or both. The correction rate in the treatment group is 23.8%, fourteen times higher than the 1.6% in the control group. This response is asymmetric: corrections are primarily made by sellers, even when only buyers receive letters, providing evidence that some firms induce a change in their partners’ tax reporting behavior. Spillover effects extend to transactions not listed in the letters, including those involving other trading partners. The intervention also results in sustained improvements in reporting behavior over subsequent months.

Tax audits are essential for governments to raise revenue but they can create economic distortions. To avoid the financial burden of an audit, firms may remain small, move to the informal sector, or shut down. Leveraging detailed administrative tax data from the Ugandan Revenue Authority (URA), a novel linked survey, and a regression discontinuity design (RDD), we show that audits have a dual negative effect in our context: They reduce the tax revenue collected from audited firms \textitand impose large economic distortions. Audited firms are 11 percentage points (p.p.) likelier to shut down, and those that remain operational reduce their output. The former result is driven by firms that must pay substantial back taxes and the latter by firms that believe they are likely to be audited again soon. Back-of-the-envelope calculations indicate that the overall revenue collected from audited firms declines. The total wage bill loss induced by the audits is equivalent to 0.2-0.6% of the total wage bill of the formal economy at baseline, suggesting that the auditing process potentially imposes a large distortion to the Ugandan economy.

Making use of a rich administrative dataset on Ugandan firms’ tax filings covering the period 2013–2021, this paper investigates the impact of tax audits on voluntary compliance, contrasting the effect of one versus multiple audits. Using a matched Difference-in-Differences approach with similar unaudited firms as controls, and a stacked design to address staggered treatment timing, the analysis shows that among firms that consistently file taxes over the study period, audits induce higher value-added tax (VAT) liabilities. Crucially, this is \textitentirely driven by firms receiving multiple audits, underlining the importance of repeated interactions with the tax authority for fostering compliance among this set of taxpayers.

Based on a randomized experiment with dairy cooperatives in western Uganda, we provide causal evidence that SMS-based digital receipts for daily milk deliveries improve accountability, product quality, and delivery frequency. In a context of weak monitoring and imperfect transaction information, these messages allow smallholder farmers to better observe the behavior of intermediaries in the supply chain. The intervention effects vary with the intensity of information frictions. Among farmers facing a high level of information friction (i.e., those relying on intermediaries to transport the product), the intervention increased their detection of discrepancies and encouraged switching from dishonest intermediaries. Farmers in low-friction settings (those who deliver their own product) delivered milk more frequently; the likely mechanism was the nudge provided to these farmers when they received messages both on days that they made deliveries and days that they did not. We also found that the intervention increased milk quality for both self-deliverers and farmers using transporters. Overall, we provide evidence that simple digital tools can reduce information asymmetry and strengthen accountability in smallholder supply chains.