Tax avoidance by multinational corporations: an empirical analysis based on firm-level data
|Author:||Sarah Godar (5.10.2022)|
|Year:||2022 - winter|
|Leaders:|| doc. Petr Janský Ph.D.
|Work type:|| Dissertations
|Awards and prizes:|
|Abstract:||In this thesis, I use confidential firm-level data from the Microdatabase Direct Investment (MiDi database) provided by the Deutsche Bundesbank to analyze tax avoidance by German MNCs. While such data has frequently been used in tax-avoidance research, it has yet to be employed to derive macro-level estimates of tax avoidance by Germany-based MNCs. My MiDi-based research includes an estimation of the scale of profit shifting by German affiliates of foreign MNCs and related tax-revenue losses (Chapter 1), as well as an investigation of the tax-haven use and distribution of profits and economic activity of MNCs headquartered in Germany (Chapter 2). Finally, in collaboration with several co-authors, I investigate relatively new micro data on the global tax payments and activities of multinational corporations, voluntarily published by individual MNCs following the implementation of the new CbCR standard (Chapter 3).
I employ different methodological approaches depending on the quality of the data and the research focus of each chapter. In the first chapter, I employ a standard microeconometric approach to identify
profit-shifting and estimate the semi-elasticities of MNCs’ profits with regard to changes in tax incentive variables. I find that the profits of German affiliates are highly sensitive to foreign tax rate changes. The semielasticity is higher when at least one investor is located in a tax haven and is not significant when a company has never had a tax-haven investor. The estimated effects are used to extrapolate aggregate revenue losses, which range between EUR 1.5 and 5.8 billion in 2016.
In the second and third chapters, the research methodology is descriptive. The research focus of the second chapter is mainly on examining the allocation of the profits and economic activity of MNCs headquartered in Germany, at a time when German CbCR data was not yet accessible. Based on a sample of German parent companies and their foreign affiliates, my co-author Petr Janský and I, analyze to what extent the location of the MNCs’ profits is misaligned with the location of their economic activities in terms of measured in terms of the number of employees, assets, and turnover. The descriptive methodology does not allow for the identification of profit shifting but provides some relevant insights based on a previously unexplored dataset. These include the relative stability of total misaligned profits over time and the relatively moderate overall scale of misaligned profits.
The third chapter, co-authored by Giulia Aliprandi, Tommaso Faccio and Petr Janský, relies on a relatively small but original dataset of voluntarily published company-level CbCRs. We assess the value added and limitations of qualitative and quantitative information provided in the reports also based on comparison to individual MNCs’ annual financial reports and aggregate CbCR data. We find that early publishers of CbCRs do not double-count profits by including intra-company dividends and that some correct their profits for equity-accounted participation results. We further provide a tentative framework to evaluate tax risk indicators across sample MNCs and assess their potential overall tax aggressiveness even in the absence of a clear identification of profit shifting.
The results of this thesis suggest that cross-border tax avoidance by MNCs is substantial but that significant heterogeneity exists among firms. First, MNCs seem to shift less profit out of their headquarter jurisdictions than between foreign affiliates. Second, larger Germany-based affiliates are more likely to have ownership links to tax havens — and firms with ownership links to tax havens seem to engage more extensively in profit shifting. Third, MNCs that are more tax transparent avoid certain reporting problems identified in aggregate CbCR statistics and score low on indicators of tax aggressiveness.