StuW Sonderheft NeSt 2025

StuW Sonderheft NeSt 2025

Abhandlungen – Steuerwissenschaften

S71

Koch / Scheider – The Known Unknown: Tax Avoidance by European Multinationals

3. Data and Methodology We base our empirical analysis on financial statement informa tion for a sample of large European publicly listed firms over the period 2011 to 2020. Starting point for our sample selection is the universe of all firms currently listed 2 in the Stoxx Europe 600. This index comprises the 600 largest listed firms in Europe (including Switzerland and UK), which are ranked and selected based on the market capitalization of free float. It currently comprises firms from 17 different European countries. 3 The Stoxx Europe 600 covers approximately 90 percent of the over all market capitalization of European stock markets ( STOXX , 2017). Relying on European listed firms allows us to consider a sample from different jurisdictions, which base their financial statements on the same accounting regulations. 4 We start our sample selection with a total of 6,000 firm-year observations, referring to 600 firms over the period 2011 to 2020. We exclude firm-years with missing tax reconciliation data as well as all banks, as firms from this industry may be subject to special accounting standards or tax regulations. In line with prior research, we exclude loss observations in order to avoid problems with interpreting ETRs ( Drake et al., 2019; Schwab et al., 2022). We also remove influential outliers in ETRs or ETR items from our sample. Following Schwab et al. (2022), we regard observations with effective tax rates smaller than zero or larger than one as influential outlier. Similarly, we disregard observations which report values smaller than minus one or larger than one for at least one effective tax rate item. Finally, we drop observations with missing information on to tal assets or return on assets. Our final sample comprises 3,985 firm-year observations referring to 509 different firms. Table 1 summarizes the sample selection process in detail. We provide information on the geographic and industry-level breakdown of our sample in Appendix 1 and 2.

Blouin & Robinson (2020) argue that this problem is inherent in many of the typically used datasets, including the U.S. BEA data and the country-by-country reporting data. It is also pre sent in the ORBIS data. They suggest that profits should be ad justed for equity income because otherwise estimates of inter national profit shifting are likely to be overstated. In principle, however, dividends and equity income are not directly reported in many of these datasets. 1 Blouin & Robinson (2020) also show that the measurement error from not properly correcting for double counting can be large. They estimate that their pro posed correction reduces the estimate of the U.S. fiscal impact of base erosion and profit shifting from 30 to 45 percent to 4 to 8 percent of corporate tax revenues. In addition to these data-related concerns about the accuracy of previous studies, these studies have limitations when it comes to drawing conclusions about the extent of profit shift ing by European multinationals in the post-BEPS era. First, many of the studies mentioned above are based on a broad set of countries or on information from non-European countries. However, we know from previous research that tax avoidance by European and U.S. multinationals and its evolution over time are different ( Thomsen & Watrin , 2018). Second, many studies refer to the pre-BEPS period. We expect that MNEs are likely to have adjusted their tax avoidance policies in response to more restrictive tax regulations. In addition, stricter tax en forcement as well as media coverage of tax shelters may have changed firms ’ attitudes toward tax avoidance ( Dyreng & Han lon , 2021). Our study is one of the first to quantify profit shifting by Eu ropean multinationals over a time period that includes years after the implementation of anti-BEPS measures. Our analysis is based on a panel of consolidated financial information for a sample of listed European multinationals over the period 2011 – 2020 (see Section 3 for details). The use of consolidated data allows us to avoid many of the problems mentioned above. We avoid problems with the coverage of financial information inherent in the use of unconsolidated financial statement data. Consolidated data reflect the tax burden of all consolidated af filiates, regardless of local disclosure requirements and enforce ment. The double counting problem of foreign profits is also avoided by the consolidation mechanism. Our approach is inspired by studies that assess trends in tax avoidance by analyzing the evolution of effective tax rates over time ( Dyreng et al., 2008; Markle & Shackleford , 2012; Thom sen & Watrin , 2018). However, we acknowledge the findings of recent literature that effective tax rates do not only reflect the results of international profit shifting, but are also affected by other influences, such as tax rate changes, the use of local tax credits, or the tax-efficient use of tax losses ( Drake et al., 2020; Schwab et al., 2022). We analyze hand-collected information from the IFRS tax reconciliations to separate the effects of these various influences and isolate the impact of using international tax rate differences on GAAP effective tax rates. This approach therefore allows us to assess how the overall level of profit shift ing has evolved over time. It also allows us to examine whether European multinationals have replaced international profit shifting with other tax planning schemes or whether they are actually avoiding less tax in the post-BEPS era.

Table 1: Sample Selection

600 Firms over 10 Years

6,000 ./. 360 ./. 655 ./. 458 ./. 421

Excluding Banks

Excluding Missing Tax Reconciliation Data

Excluding Loss Years

Excluding Influential Outliers for ETR and ETR Items

Excluding Missing Information on total assets or return on assets

./. 121

Final sample

./. 3,985

This table summarizes the sample selection process.

Information on effective tax rates and effective tax rate items are hand-collected from the tax reconciliations, which are, ac cording to IAS 12, reported as a mandatory item in the income tax note of IFRS financial statements. The tax reconciliations present a tabularized reconciliation of the theoretical tax rate,

1 For example, ORBIS does not allow for a direct splitting of financial rev enues into dividend income and interest income. 2 We refer to the index composition as of November 2021. 3 The index composition is reviewed on a quarterly basis, which may also affect the distribution across countries ( STOXX , 2017). The country dis tribution is shown in Appendix 1. 4 Regulation (EC) No. 1606/2002 governs the reporting of financial state ments in accordance with IFRS for all EU listed firms. The UK has adopted this regulation. Swiss firms have the choice between IFRS, US GAAP and Swiss GAAP, but mainly select IFRS.