Francis Bloch* and Gabrielle Demange*
This article was originally published in the September 2021 edition of the 5 papers…in 5 minutes.
The rules for taxing the profits of multinational corporations were devised in the 1920s and are thus clearly not adapted to the digital economy; in particular, they allow GAFAM largely to escape paying any tax on their profits. To address this problem, several countries, including France, have proposed the unilateral imposition of a specific tax on the revenues of digital platforms. At the multilateral level, following recent announcements by the Biden administration, more than 130 countries seem close to agreeing on two new international taxation reforms (1). The first, called Pillar One, gives countries in which multinational corporations operate the right to tax a fraction of their profits. Pillar Two introduces a minimum tax rate of 15%, authorising the country hosting the head office of such a business to impose tax on any foreign profit taxed at less than 15%.
In this article, Francis Bloch and Gabrielle Demange analyse the effect of the differences in tax rates imposed under Pillar One on the behaviour of monopoly platforms. They focus on the network externalities, both positive and negative, created by the participation of actors in different countries on those same platforms. They compare two profit-sharing rules: the first is based on the truthful accounting of profits made by the platforms in each country (referred to as Separate Accounting); the second, which applies where fiscal authorities have no access to profits allocates profits on the basis of a distribution key, such as the number of users in each country (referred to as Formula Apportionment).
Their initial results show that even in the absence of a transfer mechanism such as the payment of royalties or interest on loans among subsidiaries, a digital platform company can transfer part of its profit to countries with low tax rates. Under Separate Accounting, it can exploit network externalities by manipulating demand and profits in the two countries; under Formula Apportionment itadjusts the number of users to influence the distribution key. This company response to different tax rates leads to distortions, which are analysed and reported in a second set of results. Here, the authors distinguish between the direct and indirect effects of an increase in a country’s tax rate. The direct effect is calculated assuming that the quantities remain fixed in the other country. The indirect effect arises when quantities in the other country are adjusted because of network externalities. In Separate Accounting, the sign of the direct effect depends on the sign of the network externalities: if the latter are positive, then the platform company increases the quantity in the high-tax country, while it lowers it when the externalities are negative. Under Formula Apportionment, the platform always has an interest in lowering the number of users in a high-tax country. Thus, we see that when externalities are positive, the effect of a tax increase on prices and quantities are the opposite under the two systems: prices fall in high-tax countries under Separate Accounting but increase under Formula Apportionment.
Bloch and Demange conduct a number of simulations in order to compare prices, profits and fiscal revenues in the two different profit-splitting systems. Their simulations show that distortions are higher under the Formula Apportionment than under Separate Accounting. Moreover, the two countries disagree in the choice of the profit-splitting regime: given fixed tax rates, the high-tax country has higher fiscal revenues under Separate accounting whereas the low-tax country has higher fiscal revenues under Formula Apportionment
(1) Cf the authors’ analysis – Le Monde editorial, 27 Juillet 2021 : « L’accord de Venise sur la taxation des multinationales reste à négocier »
Original title of the article: Profit-splitting rules and the taxation of multinational Internet platforms
Published in: International Tax and Public Finance 28(4), 855-889 (2021)
Available at: https://hal.archives-ouvertes.fr/hal-02502468/
* PSE Member