Implementing a value added tax (VAT) system to simultaneously increase public revenues, increase economic efficiency and reduce inequalities is a significant challenge for developing countries. The question of the distributional impact of VAT design has received much attention in the literature. While VAT is a general equilibrium policy, its impact has been primarily considered in partial equilibrium contexts. However, the VAT is not only paid by final consumers in developing countries. VAT becomes a reporting burden for producers with exemptions and a financial burden if refunds of VAT credits are not timely. In this paper, we use a two-step modelling procedure —a computable general equilibrium (CGE), followed by a micro-simulation—to analyse the distributional and economic impact of various VAT designs for Niger. Our simulations show that while a flat rate is best for economic efficiency, a higher statutory VAT rate (at around five percentage points) with exemptions for staple foods, has the greatest potential for poverty reduction. When the two objectives are combined, a multiple rate is the best option if VAT credits are refunded. By using a disaggregated macro–micro framework, we illustrate the importance of capturing the specificities of VAT design to measure its distributional and economic impact.