Conference Agenda

This study investigates China's recent Value-Added Tax (VAT) immediate refund reform aimed at improving production efficiency. Utilizing detailed firm-level data from 2009 to 2019, we explore how allowing immediate refunds of excess VAT input credit affects firm growth, and how these impacts propagate through supply chains. Our findings show that immediate refunds spur firm growth through two channels: (1) by providing liquidity to firms previously burdened by persistent excess VAT credit, and (2) by reducing the marginal cost of any refund-generating activities such as large investment and input expansions. These direct effects also result in positive growth spillovers to upstream industries leading to a likely net-tax revenue increase. We also show that the benefits of statutory immediate refund mandates can be limited by the local administrative environment.

Missing trader fraud is widely viewed as a major source of VAT revenue loss in the EU, yet existing evidence remains limited and largely based on indirect or aggregate measures. We use administrative audit data covering firm-to-firm cross-border transactions across all EU Member States to study this fraud. We first document new facts on its extent and nature, showing that detected fraud is concentrated across a few borders in Eastern and Southern Europe, in the retail sector, and among small rapidly expanding firms. We then examine the effectiveness of two policy instruments, tax audits and the reverse charge mechanism (RCM) in curbing this fraud. Event-study estimates show that tax audits have substantial effects: following detection, many missing traders exit while surviving firms contract, with similar dynamics observed among exporting trade partners. In contrast, RCM lead to substitution of fraud to neighboring non-treated sub-sectors, limiting overall deterrence effects and reducing its impact.

The indirect tax burdens are generally assumed to be shifted fully or predominantly to buyers. Does the same logic apply to export VAT rebates? This study examines the tax incidence of export VAT rebates both empirically and theoretically. We take advantage of China's 2004 export VAT rebate as a natural experiment, and estimate the pass-through rate by employing the difference-in-differences approach. Combining the comprehensive product-level data from China Customs (2003-2004) and detailed export rebate rate information, we find that exporters bear 100% of the export VAT rebate burden. We further show that this reduction of export VAT rebates significantly lowers export quantities and values. We then develop a theoretical model examining the tax incidence of export rebates from one country. This model rationalizes our empirical finding that the burden cannot be shifted to foreign buyers. Our paper provides an insight into the distributional effects of export tax rebate policies.

Randomized response techniques (RRTs) are commonly used to gather sensitive information, such as tax evasion, under the assumption that privacy protection removes social desirability bias. Using a survey of 1,742 German firms, we measured VAT evasion via RRT and competitor evasion without RRT. Our results show that social desirability bias leads to a violation of the RRT protocol (reporting “no” personal evasion) and, for consistency, a report of zero competitor evasion. This creates a 9.2 percentage point difference in evasion estimates, unlikely reflecting real differences. The discrepancy is particularly pronounced among firms sharing an email address and those in professional services, suggesting that they have stronger reputational concerns. Additional validation using realistic entrepreneurs' responses generated with Llama-3.2 reveals comparable effects (-7.0 percentage points), demonstrating that desirability-driven response behavior emerges in both human and artificial data. Overall, while valuable, privacy-enhancing survey methods cannot fully eliminate reputational and self-image-driven response distortions.