Refunds are an essential feature of well-functioning VAT systems and take up a sizeable portion of government spending. In South Africa, refunds amount to 50 percent of gross VAT collection, representing a substantial transfer from the government to taxpayers that has to occur at relatively high frequency, often monthly. We show that delays in these refund payments reduce domestic investment, especially by small firms. We use administrative data to provide extensive evidence that firms respond to incentives created by delays and denials of refunds. We exploit a change in the attitude towards refunds of the South African Revenue Authority to quantify these effects. We find that approximately halving the audit rates of refund-claiming returns and speeding up their processing increased investment by 31 percent and output by 24 percent.

We study the impact of adopting the value-added tax (VAT) on countries' tax revenue performance. Applying an event-study design and using panel data for 74 countries that introduced a VAT between 1986 and 2015, we estimate that adopting the VAT increases the tax-to-GDP ratio by about 1.5 percentage points on average. The effect is larger, up to 2 points of GDP, for low- and middle-income countries, in particular those located in Africa and the Middle East. The increase in VAT revenue more than offsets the decline in other indirect taxes—particularly trade taxes—consistent with VAT adoption replacing more distortionary sources of revenue, while income tax revenue remains broadly unchanged following adoption.

If collected using an invoice-credit system, the VAT is often considered to be tax on consumption that is neutral with regard to input sourcing as taxes on inputs get reimbursed. Yet taxes need to be charged, remitted and reimbursed for transactions along the value chain, resulting in compliance cost and making tax administration non-trivial in particular in the context of cross-border trade. EU countries have therefore partly shifted the responsibility to remit VAT from seller to buyer in the B2B context. We explore the effects on input sourcing using a comprehensive database of the VAT returns of the universe of German firms. The results point at significant effects of changing the mode of VAT collection on firms. Our findings support the view that VAT enforcement along the value chain distorts input sourcing.

In this paper I use Czech administrative tax data from 2015-2023 to study whether there are differences between compliance behavior of firms and self-employed individuals at the VAT registration threshold. I apply the novel diffuse bunching framework of Anagol et al. (2025) to analyze whether bunching responses are motivated by a tax notch, fixed costs of VAT registration or optimization frictions. I find that their underlying revenue elasticity with respect to the tax rate is similar (around 0.24), suggesting comparable responses to tax incentives. However, the two groups differ substantially in their perceived fixed costs of VAT registration and their ability to choose optimal revenue. I find that self-employed individuals face 3 times higher fixed costs of registering to the VAT. I also find that firms exhibit much lower optimization frictions than self-employed individuals.