Tax haven investments for Profit-Shifting: evidence from Spanish multinationals

ABSTRACT

Tax havens may play a key role in the Profit-Shifting (P-S) activity of Multinational Companies (MNCs), since, among other characteristics, they are the territories with the most beneficial taxes for foreign investors. This article shows that Spanish MNCs facing higher tax rates in non-tax havens, and therefore those that stand to gain the most from P-S, are more likely to invest in tax havens. This outcome is robust to at least two different tax haven lists and various definitions of the non-haven tax rate. The size of the MNCs’ activity, their use of intangible assets, and belonging to the Ibex 35 stock index also positively affect the probability of investing in tax havens. By economic sectors, once the endogeneity problem is controlled for, the incentive is greater for manufacturing than for service firms, but it is especially high for financial firms. Additionally, while non-haven tax rates positively influence the number of different tax havens used by firms, they have no effect on the number of affiliates located within them. Finally, the article estimates that Spanish MNCs have been able to save about 4 billion euros per year in Corporate Income Tax in the period 2013–2018 as a result of P-S.

KEYWORDS:

• Multinationals
• tax haven investments
• Profit-Shifting
• probability model

JEL CLASSIFICATION:

• F23
• H25
• H26
• H87

Acknowledgements

The authors would like to thank the reviewers for their comments and suggestions. A version of this article was presented at the Conference on “Tax Evasion or Avoidance and Tax Havens, from the Nineteenth Century to the Present Day”, 24th and 25th June 2021, University of Lausanne.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available on request from the authors. The data are not publicly available due to commercial restrictions.

Notes

¹ Regarding the magnitude of P-S, there are some papers that have recently taken advantage of the Country-by-Country reports to get more accurate estimates. See, for instance, Bratta, Santomartino, and Acciari (2021), Clausing, Saez, and Zucman (2020), Clausing (2020a and Clausing 2020b), Fuest, Hugger, and Neumeier (2021) and Fuest et al. (2022).

² Desai, Foley, and Hines (2006) obtained a negative effect of taxes on investment in the smaller tax havens. They took this contradictory result to mean that the smaller tax havens were being used by US MNCs to delay US taxation of lightly-taxed foreign income. This last result might be related to the International Tax System applied by the US. Until the 2017 Tax Cuts and Jobs Act (TCJA), the US applied a worldwide tax system, which meant that companies had to pay the difference between domestic and foreign taxes when profits were repatriated. Only MNCs headquartered in countries applying a territorial tax system, like most OECD MNCs, can benefit from reduced foreign taxes at the end of the day.

³ Spain applies the exemption method for the correction of double international taxation with those territories that Spanish law does not consider to be tax havens and with which a Double Taxation Treaty is in force or which have a statutory Corporate Income Tax rate of 10% or more. Spain also applies the exemption method with Gibraltar. Additionally, we assume that there is a certain level of economic activity in the territories where the Spanish companies operate. Otherwise, according to the special regime of Controlled Foreign Corporations applied by Spanish Corporate Tax Law, profits are taxed in Spain directly, without any tax benefit from the P-S activity.

⁴ According to the Bureau van Dijk, very large companies are those with turnover higher than €100 million, total assets higher than €200 million or number of employees higher than 1,000; and large companies are those with turnover higher than €10 million, total assets higher than €20 million or number of employees higher than 150.

⁵ German MNCs are firms who hold shares or voting rights of 10% or more in a company with a balance sheet total of more than €3 million. The percentage of German companies that have any tax haven affiliate is 82% for large MNCs subject to Country-by-Country reporting in 2016 and 2017 (Fuest, Hugger, and Neumeier 2021).

⁶ Gumpert, Hines, and Schnitzer (2016) took the number of employees as a proxy for the real profits of the individual affiliate (adjusted by the participation of the parent in the affiliate). Instead, we weight the tax rate of each country by the number of affiliates within it, because there are a lot of missing data in the Sabi database for financial variables.

⁷ It is assumed that missing data are randomly distributed among companies.

⁸ We took the companies belonging to the Ibex 35 index from 2013 to 2018, most of which are active in the service and the financial sectors.

⁹ Different from Gumpert, Hines, and Schnitzer (2016), we ruled out the estimation of firm fixed-effect models due to the pooled structure of our data, which an increasing number of companies over time; and the limited number of firms changing their status of having a tax haven affiliate or not, which reduces the estimates to 8.38% of the sampled firms when applying firm fixed-effects. However, we consider that all important factors affecting the opportunity of firms to accomplish P-S are controlled for by the model variables and the industry dummies. Logit and Probit models were also estimated, and the sign and statistical significance for all the model variables remain very similar to those presented in the following section.

¹⁰ Then, for every MNC, one year is lost from the sample, so that the sample is limited to the period 2014–2018. Moreover, those MNCs with available information from 2018 on are eliminated from the sample.

¹¹ Thus, it seems that the tax rate variable is exogenous and does not need to be instrumented. Nevertheless, when instrumented, the F statistic is significant, which means that the instrument is valid.

¹² We did not consider the Spanish CIT rate in the previous sections because, although the probability of investing in tax havens may also be influenced by the CIT rate in the MNC residence country, this effect cannot be estimated due to lack of variation (Gumpert, Hines, and Schnitzer 2016).

¹³ The tax haven list taken in this section is again the extended Hines and Rice (1994) list.

¹⁴ If we approximate the share of the global Corporate Income Tax base for each country with its GDP, we can estimate that, of the €4 billion saved per year, €660,000 thousand would correspond to P-S from Spain to abroad and, therefore, the rest to P-S through affiliates established in other countries. The latter figure would also approximate the maximum revenue loss for Spain as a result of that activity (Table 12), i.e. the loss if the tax rate of the benefit recipient country were zero.

¹⁵ See https://www.oecd.org/tax/beps/beps-actions/action1/. For a recent assessment of the effects of implementing the Pillars (including revenue effects), see IMF (2022).

Additional information

Funding

The work was supported by the Department of Science, University and Knowledge Society of the Government of Aragon [project S23_20R].