Value-added tax (VAT) registration thresholds are among the most common size-based

regulations worldwide, yet existing evidence has focused mainly on firms’ avoidance behavior below these thresholds. This paper studies what happens when small firms enter the VAT regime. We use newly available German administrative tax data covering the near-universeof firms and exploit the €17,500 VAT registration threshold. To comapre crossing and noncrossing firms we combine a dynamic difference-in-discontinuities design with a donut RD approach.

Due to data protection reasons we cannot disclose the full paper publicly.

Many countries use a reverse charge mechanism (RC) in the value added tax (VAT) system to combat tax evasion in specific high-risk sectors. The RC shifts the liability to remit VAT from the seller to the buyer. We study the adoption of RC in the construction sector in Finland in 2011 using tax return data on the universe of Finnish firms. Using a difference-in-differences design, we find that reported net VAT liabilities in the construction sector increased by 5%.

According to our results, changing the remittance policy decreased VAT evasion by small subcontractors that provide services for large firms. Using a theoretical model, we show that reverse charge increases tax revenue relative to conventional VAT unless downstream firms increase tax evasion beyond the level previously undertaken by upstream firms, a response that requires a high degree of non-compliance.

We examine the economic consequences of delayed Value Added Tax (VAT) refunds in developing countries, using a novel administrative dataset covering the universe of VAT refund claims in Zambia. We document severe delays in refund processing, with firms waiting, on average, nearly 700 days for reimbursement—over 20 times the statutory processing period. Merging refund records with the universe of firms' VAT and Corporate Income Tax (CIT) filings, we show that prolonged refund delays have substantial adverse effects on firm performance, reducing sales, taxable purchases, profits, investment, and employment. To address the endogeneity of refund delays, we exploit plausibly exogenous variation in administrative congestion using a leave-one-out measure of the average delay experienced by other firms entering the refund system at the same time. The instrumental-variable estimates indicate that the economic costs of refund delays are large and persistent, consistent with firms facing binding working-capital constraints while awaiting reimbursement.

Institutional change tends to arise from "reform journeys", where incremental changes—rather than one-off events—lead to long-run progress. This paper studies the evolution of VAT misreporting in Kenya during a period of sustained administrative reform using firm-to-firm transaction data from 2016 to 2023. We document widespread underreporting of VAT liabilities, especially on the extensive margin, and a sharp decline in this behaviour over time. Using a fixed-effects regression model that allocates reporting discrepancies to buyers or sellers, we show how systematic misreporting varies across firm size, sectors, and tax offices. Eliminating firms’ misreporting would have increased KRA's tax revenue by KSH 597 billion (40% of VAT payable). The negative revenue impact has decreased significantly over time, highlighting improvements in KRA's tax capacity during a sustained reform period.