Offshoring in the European Union: A Study of the Evolution of the Tax Burden

One of the most serious effects of offshoring is tax avoidance, which harms the economies of the affected regions. In an attempt to eradicate tax avoidance, the EU seeks to establish tax harmonization across its Member States. Based on data for 2006–2014, this study analyzes the historical evolution and current trends of a convergence or divergence of the tax burden for 15 EU Member States. The effective tax rate was used to assess the tax burden. This study used a novel approach to analyze the tax burden and conducted a cluster analysis to examine changes in the effective tax rates between 2006 and 2014. The results imply that when the economy prospers, effective tax rates tend to converge. In contrast, during periods of economic downturns, effective tax rates diverge. This divergence occurs because of differences in Member States’ tax policies that reflect the various strategies that are adopted by different Member States to combat economic crises. Therefore, the tax harmonization criteria that were established by the EU are relegated to the background and offshoring is encouraged.


Introduction
Since Kaplan's (1975) study regarding effective corporate tax rates, scholars' interest in the tax burden has increased because of different tax policies that have been adopted in successive macroeconomic environments, both locally and globally. The tax burden is such a popular topic, that certain scholars have written reviews on tax research (Hanlon & Heitzman, 2010). Different research streams focus on different aspects of the tax burden. Scholars have studied the tax burden from a macroeconomic perspective (Loretz, 2007) and a microeconomic perspective using data from comtions to ensure countries work more closely together to eliminate controversial legal loopholes that enable tax avoidance.
This new Council Directive seeks to minimize offshoring and ensure that companies comply with their tax obligations in the country where they oper-ate, thereby fulfilling their social duty, rather than simply moving their taxable income to countries with lower tax rates. The EU's goal is to achieve full harmonization to prevent the implementation of aggressive tax policies that lead to tax avoidance by tax-paying companies.
In accordance with the goal of Council Directive (EU) 2016/1164 of 12 July 2016, the European Commission recently heavily fined Apple for taking advantage of low tax rates in Ireland. In doing so, the European Commission sent a warning to other large firms that tax engineering would not stop them having to pay the taxes they owe.
Notably, the aforementioned Council Directive was approved only last year. Therefore, the issues that are addressed in this study are highly topical, which justifies its relevance. This study used empirical data to investigate whether the primary EU Member States (including the United Kingdom) differ significantly in terms of their tax burden. Accordingly, the effective tax rate (ETR) of 15 EU countries was analyzed. The ETR was used because it is the most widely used indicator to measure a country's tax burden (Armstrong, Blouin, & Larcker, 2012;Fairfield & Jorratt De Luis, 2016;Kaplan, 1975). Our primary goal was to study how these differences evolved over time to determine whether there is a tendency towards convergence that would lead to tax harmonization and subsequently, the absence of undesirable offshoring to countries with lower tax rates.
This study contributes to extant literature in several ways. First, this study offers a current analysis of the historical evolution and current trends regarding a convergence or divergence of the tax burden across Europe that results from tax policies that have been applied by EU Member States. Second, the number of countries included in the analysis (15) is considerably larger than the number of countries that have been analyzed in prior studies that compared tax systems across European countries. Therefore, this study provides a substantially broader perspective of the current situation. Third and most importantly, the study's primary contribution to extant literature is the application of cluster analysis, a novel technique for studying the tax burden. Cluster analysis is a method that is commonly used in other fields, such as medicine and the geosciences, and allows us to group countries based on Offshoring in the European Union: a Study of the Evolution of the Tax Burden This work is licensed under a Creative Commons Attribution 4.0 International License. the similarity of their ETRs; therefore, it should reveal how tax harmonization has evolved over time.
The ETR was chosen as an indicator of the tax burden in each country for this study's empirical analysis based on a rigorous literature review. From a methodological perspective, we selected the sample to avoid biases and justify the procedure that we used. For example, outliers were removed because they might have distorted the mean. In addition, non-representative data were removed from the data set.
The primary result is that during periods of economic prosperity, there is a tendency towards a common tax policy. Accordingly, the tax burden tends to converge. In contrast, during periods of economic downturn, an absence of convergence can be observed and each country applies its own tax policy. This situation results in divergence in the tax burden.
The paper is organized as follows. Section 2 presents the theoretical framework, reviews extant literature and states the hypothesis. Section 3 describes the method, sample characteristics, and variables that are used in the empirical study. Section 4 presents and discusses the results of the analysis. Finally, Section 5 discusses the conclusions of the study.

Literature review and hypotheses
The fundamental goal of studies that compare tax burdens in different regions is to determine whether tax burdens differ, which may affect decisions regarding company location and therefore, may affect the conditions necessary for a free market. Notable studies include those conducted by Chennells and Griffith  The authors determined that the countries with the greatest spread between the two rates were Germany, Italy, and Spain. Jacobs and Spengel (2000) compared the ETRs of companies located in France, Germany, the Netherlands, the UK, and the US over 10 years. These scholars used an innovative measure of a specific ETR that was calculated as a percentage of profit and indicated how much profitability each company lost because of taxes. These scholars determined that tax burdens varied across different sectors within the same country and across different countries within the same sector.
They reported that the UK had the lowest tax burden, followed by the US and the Netherlands. France and Germany had the highest ETRs.
Overesch and Rincke (2011)  Offshoring for tax reasons is an issue that greatly concerns governments. In response to this problem, numerous scholars have studied variations in the tax burdens of different countries and attempted to identify significant differences that may lead to offshoring (Buijink et al. 2002;Lisowsky, 2010).
Lisowsky (2010)  The authors report that financially constrained multinationals shift 20% less income from the US to foreign countries than unconstrained multinationals.
Another notable study, Buijink et al. (2002), analyzed tax policies of 15 EU Member States between 1990 and 1996. The author sought to identify substantial differences in tax burdens (in terms of the ETR) that might influence company location decisions by offering competitive advantages that distort free competition. The authors determined that tax incentives differed significantly among countries and caused greater variations in ETRs than in STRs. Buijink et al. 's (2002) study offers a starting point for this study because of its similarities in terms of geographical coverage and sample characteristics. Although certain aspects of their study differ from the current study, both studies sought to observe differences in tax burdens.
Based on the literature review and the scope of the present study, the two following hypotheses are proposed: Proposition for the first hypothesis: Tax rates are identical across the 15 EU countries that are analyzed in this study. Therefore, Null hypothesis H 0 : ETRs and STRs are the same among the 15 EU countries that are analyzed in this study.

Tax preference measures
In the study that acted as the precursor to tax research, Kaplan (1975)

Research design
To determine whether significant differences exist between the means of the tax burdens in the countries that were analyzed in this study, we performed ANOVA and a robust test of differences between the means of the STRs and the ETRs for each country. Any significant differences that were revealed by the analysis were then examined in greater detail to analyze these differences and their evolution over time. Therefore, it was possible to assess the tax harmonization policy of the EU.
A cluster analysis was performed. Cluster analysis is a multivariate statistical method that is used to group objects based on their similarity and is defined as the distance between objects (Loster, 2013). The distance is calculated based on the distance matrix, which is constructed using the squared Euclidean distance, as per the following formula. The analysis was conducted in SPSS, which forms clusters of objects according to the squared Euclidean distances among those objects. Cluster analysis is commonly used to study groupings in other disciplines, such as geoscience (Asante & Kreamer, 2015) and medicine (Sonğur & Top, 2016). However, the use of cluster analysis is a novel approach to analyze ETR.
Although Regis, Cuestas, and Chen (2015) and Chen, Cuestas, and Regis (2016) conducted their studies using cluster analysis, they used a different technique and examined only the STR. Table 2 provides the results of the ANOVA that was conducted in SPSS. Based on the data in Table 2, the null hypothesis is rejected. Therefore, significant differences exist in the mean values for STR and ETR among the EU15 countries. This result is consistent with the results that were reported by Buijink et al. (2002), which analyzed the evolution of differences and tax harmonization between 1990 and 1996.

Robust test of differences between means
The robust test of differences between means indicates that the differences between means are significant for almost all pairs of countries for STR and ETR. Tables 3 and 4 present the results of the robust test of differences of means between pairs of countries for the STR and ETR across all years included in the sample.
The first null hypothesis, which proposed that the tax rates for all 15 countries are equal, is rejected. This

Nominal Germany Austria Belgium Denmark Spain Finland France Greece Netherlands Ireland Italy Luxembourg Portugal UK Sweden
Germany 0    To quantify the information that is illustrated by the dendrograms, the sum of the squared Euclidean dis-tances among all the countries was calculated for each year of the study. To complete the empirical analysis, we checked the results of the cluster analysis using two tests: an analysis of the standard deviations for each year and the sum of the differences between means in terms of the absolute values of the ETR. Figure 3 provides the standard deviations of the mean ETR for each year. Figure   4 provides the sum of the differences between means across countries in absolute values.    null hypothesis is rejected and the alternative hypothesis is accepted. Notably, during periods of economic prosperity, tax rates converge and during periods of recession, tax rates diverge.

Conclusions
Countries have expressed a concern regarding the need to address tax avoidance, which has led to the recent publication of the BEPS report. This study responds to the EU's desire to harmonize the tax burden across all Member States. The goal of this study was to provide evidence of whether the tax policies that were adopted between 2006 and 2014 led tax rates to converge or diverge. If tax rates diverge, companies would be inclined to pay taxes in countries with lower tax rates. This empirical study compared the STR and ETR that were imposed upon 6,249 companies located in 15 EU countries with similar economic conditions.
A robust analysis of the differences between the means indicated that differences in STR among the countries that were included in the sample were significant. Similarly, the mean ETR differed significantly among the countries in this study's sample. These results are consistent with prior studies (Abbas & Klemm, 2013;Marques & Pinho, 2014).
Next, a cluster analysis was conducted to analyze how the tax burden evolved over the study period.
Therefore, it was possible to observe whether the trend was towards harmonization or whether harmonization did not occur. The application of cluster analysis constitutes a novel approach in the study of tax burdens. Although other authors have used this technique (Chen et al., 2016;Regis et al., 2015;), they conducted their analysis from a different perspective and focused only on STR.
The results demonstrate that during the study period, the tax burden in the EU15 countries converged and diverged at different times. Interestingly, when the economy prospered, the tax burden tended to converge. Conversely, during periods of economic crisis, whether related to financial or sovereign debt problems, the differences between tax burdens of the countries were significant.
These results imply that during periods of economic prosperity, EU countries apply tax policies that approach harmonization. However, during periods of economic turmoil, each country sets its own STR and ETR according to its unique strategy to cope with the downturn. These actions indicated that the tax unification criteria that was established by the EU is ignored and companies are inclined to engage in offshoring.