This paper presents a conceptual framework for a satellite account for the measurement of the size and structure of tax evasion and avoidance and its impact on public finance. It represents a first attempt at the creation of a unifying reporting framework for all kinds of fragmented evidence on tax evasion and avoidance, departing from the adjustments made to GDP estimates in order to account for the non-observed economy, which is distinct from the undeclared economy. The paper defines the scope of the satellite account, framing the discussion within a broad definition of fraud considering the impact on both sides of the public budget, and anticipates some conceptual differences. Finally, it reviews different strands of literature in order to provide a classification of fraud phenomena, provides an outline of the tables to be compiled and indicates some of the data sources available at the international level.

We develop a dynamical model that includes not only corporate tax revenue losses due to profit shifting of multinationals, but also tax revenue collected from capital gains and dividend taxes, as well as government borrowing costs. We use country-by-country reporting data on the operations of multinationals to estimate profit shifting, alternative operationalizations of the location of investors to proxy the tax revenues from capital gains and dividend taxes, and yields on government bonds to measure the cost of borrowing. Our results show that when these indirect costs are included, the total cost of profit shifting for developing countries increases significantly, while some developed countries can offset or recover the majority of the direct costs of profit shifting. The ability of the latter to do this is, however, uneven, with, for example, most European countries losing revenues from profit shifting even after indirect effects are taken into account.

This paper investigates the association between the effectiveness of mutual agreement procedures provided for in bilateral tax treaties (MAP) and foreign direct investments by analyzing MAP statistics and distinct MAP components. Using firm-level ownership data, we find that MNEs invest more frequently in countries with good MAP policy structures and qualities (e.g., MAP components and fast dispute resolution). Vertically integrated MNEs and small parent MNEs most favor countries with several effective MAP components. Overall, we provide novel evidence on MNEs’ investment responses to effective MAP, evidence that is crucial for policymakers who aim to reduce tax disputes and double taxation.