Worst of both worlds: A case against digital services tax in Brazil

Worst of both worlds: strong reasons why the digital services tax should not be implemented in Brazil

By Maurício Barros (Partner at Gaia Silva Gaede Advogados in São Paulo, former Taxpayer-Appointed Judge at the São Paulo Taxes and Fees Court – TIT/SP (2014-2019) and a former Visiting Professor at the Getulio Vargas Foundation and at the Mackenzie Presbiteryan University) & Luiz Guilherme de Medeiros Ferreira (Tax lawyer, São Paulo and Member of the Tax Litigation Commission at the Brazilian Bar Association)

Amid the covid-19 pandemic and the imminent financial crisis of companies, Draft Bill (DB) 2358/2020, drafted by Deputy João Maia, is making its way through the Brazilian Congress. If it becomes law, it will institute a digital services tax (DST) in Brazil, like similar taxes levied in other countries.

In effect, the bill intends to institute an Intervention Contribution in the Economic Domain, referred to as CIDE-Digital in the bill, which would be levied on the gross revenue from digital services provided by “large technology companies,” with the revenue allocated entirely to the National Fund for Scientific and Technological Development (FNDCT).

Under the bill, the taxable event for the CIDE-Digital tax would occur upon receipt of gross revenue resulting from three activities: (i) online advertising, (ii) provision of digital platforms for intermediating the sale of goods and provision of services, and (iii) selling data whenever it involves users located in Brazil. The taxpayers would be legal entities, domiciled in Brazil or abroad, who earn gross revenue as a result of the activities mentioned above and who belong to an economic group that received, in the previous tax year, (i) aggregate gross revenue greater than the equivalent of BRL 3 billion; and (ii) gross revenue exceeding BRL 100 million in Brazil.

Though introduced by a single congressman and not the government’s support base in congress, the bill is already causing a major controversy among scholars and taxpayers in Brazil and has been included in the recent U.S. Trade Representative investigation with respect to Digital Services Taxes (DSTs) adopted or under consideration by countries worldwide.

The bill, however, contains a series of inconsistencies, which are explored below.

The context of the bill

The bill is inspired by the Italian and French DSTs. Given this, it is important to carefully analyze how similar or different the Brazilian business situation is to that of the countries that inspired the bill.

The expansion of digital businesses in Europe led some countries to claim that there would be an erosion of the calculation base of the Corporate Income Tax, since the whole European tax system is based on the premise that a country can only tax the profit of foreign entities, in its territory, if they come from a permanent establishment, identified by their physical presence in the destination country. Thus, the entire network of treaties between these countries inhibits taxation at the source (market country) of transactions that occur in their territory if there is no physical presence of a permanent establishment.

Faced with the possibility (made possible by digitalization) of exploiting remote markets, that is, only with a digital and not physical presence, coupled with the premise of the treaties that prevent taxation at the source of transactions in their territories, the alleged under-taxation of corporate profits in consumer countries (market) was observed.

This led to several OECD initiatives in its BEPS project. However, because these initiatives have not yet reached consensus, some countries have moved towards unilateral measures for taxation on revenue (creation of DST).

In this regard, the main difference from the situation in Brazil is that selling on the European market with a digital-only presence in the market country makes it impossible to withhold tax on any type of payment from the consumer country.

However, it must be remembered that Brazil is the country with the highest taxes on international remittances, mainly for intangibles, whether paid by individuals or legal entities. In the case of legal entities, these remittances may be subject to the federal Withholding Tax, or IRRF, at 15% or 25%; the local Services Tax, or ISS, at 2% to 5%; the state VAT or ICMS, from 18% to 25% (depending on the activity – the incidence is arguable);  the social security contributions PIS/Cofins (9.25%); the FX Tax, or IOF, at 0.38%; and even another Intervention Contribution (CIDE) at another 10%! In the case of individuals, there would be an incidence, at least, of the FX Tax (IOF), which rate (6.38%) is already higher than that of the CIDE-Digital tax itself (progressive taxation from 1% to 5%).

Assuming the same transaction (for example, payment abroad for online advertising services) between European countries and between Brazil and another country, there would be no withholding tax in the market country in the first case. In the case of Brazil being the destination country of the services, the taxation could reach up to more than 50% of the value of the transaction, considering all the taxes above.

Thus, there is a radical difference in the two scenarios that requires reflection on the real need to create a DST in Brazil. Thus, before implementing this type of tax, one must consider the real need for it in light of the Brazilian tax context and its radical differences from the European context, because Brazil should not make the mistake of simply copying a European solution (questioned even in Europe) for a problem whose real existence or dimension in Brazil has not even been seen.

Beyond this preliminary analysis as to the relevance of the creation of CIDE-Digital tax in Brazil, the proposed new tax has a series of inconsistencies and unconstitutional features that prevent its approval, as will be seen below.

Non-compliance with constitutional requirements for intervention contributions

First of all, we have seen the failure to observe two fundamental characteristics of intervention contributions according to Brazilian Constitution, namely: (i) the identification of a specific economic domain in which state intervention would take place; (ii) the compatibility of the amount paid with the specific state intervention.

The very concept of a “digital economic domain” has long been rejected by the OECD itself, which has concluded that it is impossible to isolate (“ring fence”) the digital economy to tax it as a prominent field of incidence. What was observed, in fact, is that we are facing a phenomenon of digitalization of the economy, which covers the economy in a diffuse way in its most varied aspects.

Therefore, the digital economy does not exist as a separate entity in the traditional economy, but as a growing phenomenon observed in the most varied economic segments.

In this way, digitization appears much more as a form, or medium, than as an essence, which has not been uniformly assimilated by everyone. In this context, segments such as finance (digital banks), accommodation and hospitality, transportation, advertising, retail, etc., have advanced in the digitalization process. More recently, medical, educational and even legal services have quickened their digitalization process (especially as a reaction to the pandemic).

All the examples above show the digital phenomenon as a new way for pre-existing economic categories to manifest themselves and not as an economic segment in itself. Thus, confronting the premises above with the bill, we ask: what would be the economic domain reached by the Contribution of Intervention in the Digital Economic Domain?

The taxable events completely reflect the absence of a relation with an economic domain that can be called “digital.” Roughly speaking, advertising, retail sales and data sales activities are subject to the proposed tax, if they have the special characteristic of being done digitally. However, it seems quite absurd to consider that these activities can be part of the same economic domain.

As for referability, the situation is no better. Referability means the relation between the amount collected by the tax and a specific state action that can benefit, even if indirectly, the taxable persons of the tax obligation.

First, what specific state action would that be? National technological development based on the Fund’s capitalization? To what extent would this development benefit taxpayers of that tax obligation? To the contrary, the potential national technological development would benefit the entire economy in a diffuse manner, without any particular relation to advertising, retail or data sales, which would place them in the special condition of funders of the respective fund.

In addition, the referability becomes even weaker when it is noted that the taxable person may even be a legal entity not established in Brazil. How would a foreign entity benefit from a technology development fund in the country where it is not even established? Not to mention the arguable lack of Brazilian tax jurisdiction to tax the revenues of non-residents…

Thus, whether due to the impossibility of identifying a specific economic domain, or due to the total absence of reference between the state performance funded by the contribution and its taxpayers, CIDE-Digital tax does not meet the constitutional requirements for a tax of this type.

Unwanted multiple taxation of revenue and problematic taxable events

Article 4 of the bill states that domestic or foreign companies are eligible to be taxpayers and foresees that the tax would be levied on the revenues of the indicated taxpayers with the specified activities. At this point, the multiple taxation of revenue from entities established in Brazil stands out.

This is because, if there is a physical presence in Brazil and revenue is calculated, it would already be subject to the incidence of indirect taxes such as PIS/Cofins, ISS and/or ICMS. Thus, given the dynamics presented in the bill, a domestic company would have its regular statutory revenue for the purposes of PIS/Cofins, ISS and ICMS, which would coexist with another revenue for the purposes of CIDE-Digital tax, presumed according to its own criteria.

In addition, the taxable events have some serious operational problems.

The first hypothesis is the display of online advertising to users located in Brazil. As it is dealing with online and not traditional/physical advertising, it is extremely complex to establish whether the ads are viewed only by users in Brazil or to what extent this should occur. Not even the bill’s provision for proportioning the tax calculation base at this point, so that CIDE-Digital should applies only to the “portion of gross revenue proportional to exhibitions to users located in Brazil,” in the case of advertising also displayed to users located in other countries, seems easy to apply, considering the enormous difficulties in making this calculation.

Although it is provided that users who access the digital platform on a device physically located in Brazil will be considered located in Brazil, which will be determined based on the IP address that accesses the digital platform, how will this metric be applied to the quantification of the calculation base of the DST? Will each IP that accesses the ad be considered? Each access by the same user on the same day? Accesses by the same user in a predetermined range of days? How will search platforms that might access these ads via robots be calculated?

The second taxable event (provision of digital platforms for intermediating the sale of goods and provision of services) is also problematic, as it intends to tax platforms that allow users to contact and interact with each other, with the objective of selling goods or to provide services directly between them, as long as one of them is located in Brazil. This hypothesis is apparently intended to cover marketplaces, that is, platforms where third parties buy and sell goods.

However, when stating that the purchase and sale and the provision of services must occur “directly” between users, it seems that what is being covered is any social networks through which sellers/providers can interact with potential buyers/clients, in a way that all transactions between them can be conducted and completed outside the platform (external checkout and even in the physical environment).Taking into account that social networks have their highest revenue from (online) advertising, which is the first taxable event elected by the bill, there seems to be an unsustainable bis in idem, as the same revenues could be taxed twice for the same tax: one due to online advertising itself and another due to the availability of platforms for interaction between users.

In addition, in the second taxable event there is also an enormous difficulty in separating taxable revenues, considering at least one user in Brazil, and non-taxable revenues (users abroad), both for the reasons mentioned above in regard to the online advertisement and due to the difficulty of determining what “get in touch” and “interact” in the digital world should mean: would it be enough to simply visualize someone else’s profile/offer? Exchanging messages? Closing deals?

In this case, if the deals are closed “directly” by users, how can they be discovered and accounted for for the purpose of proportionalization? Hence, the verification and the concrete quantification of the occurrence of the taxable event seems quite hard to be achieved. No less incoherent is the third taxable event (sale of data), corresponding to the transmission of data from users located in Brazil collected during the use of a digital platform or generated by these users.

This is because, in addition to the confusion with targeted advertising revenues, already mentioned above, as a contribution intended to tax large multinational technology companies, it remains to be seen whether the impact would not also reach taxpayers who have nothing to do with this universe, because data sales can occur in the simplest ways from a technological point of view (both subjective and objective), such as the marketing of mailing lists by restaurants and bars or of consumer purchase information by pharmacies and drugstores (transactions that already take place currently). Therefore, nothing would prevent huge international restaurant or drugstore chains from being taxed for activities in Brazil.

At this point, there seems to be no precise reference between potential contributors and the contribution to be created, which would also make it insubstantial from a constitutional point of view.

Other inconsistencies in the bill

The universe of potential taxpayers who would be called upon to pay the contribution also stands out: legal entities, domiciled in Brazil or abroad, who earn gross revenue as a result of the activities mentioned above and who belong to an economic group that received, in the previous tax year, (i) aggregate gross revenue greater than the equivalent of BRL 3 billion; and (ii) gross revenue exceeding BRL 100 million in Brazil.

Although the text of the bill itself is not clear, its justification states that only companies that have a presence abroad would be subject to the impact, since – according to the bill – there would be “no sense in applying it to a technology company that only acts in Brazil, even if it is large, since it will not be able to transfer the profit to branches abroad.”

This point shows yet another profoundly serious constitutional problem in the project, given the unquestionable violation of the principle of non-discrimination. In addition, it is a huge disincentive to the already timid process of internationalization of Brazilian companies with domestic capital, leaving Brazil permanently as an importer and not an exporter of technology, in an affront to the fundamental objective of guaranteeing national development (article 3(II) of the Federal Constitution).

From another perspective, the tax could unbalance competition between multinational and exclusively domestic companies, which shows yet another problem of unconstitutionality in the project.

Another serious problem with CIDE-Digital tax is its potential impact on transactions already taxed by the CIDE-Technology tax, instituted by Law 10,168/2000, that levies on remittances abroad as remuneration for (i)  agreements involving the licence to use or acquisition of technical knowledge and transfer of technology (comprising trade mark and patent licences); (ii) royalties of any kind; or (iii) technical services, technical assistance, administrative assistance or “similar services”.

The purpose of CIDE-Technology is similar to that of the proposed new tax, namely “to stimulate Brazilian technological development, through programs of cooperative scientific and technological research between universities, research centers and the productive sector” (article 1), and to fund the same National Fund for Scientific and Technological Development (FNDCT). The coincidence, here, begins in the taxable events and reaches the destination of the collected funds, in a very accentuated overlap.

Finally, there is a serious operational problem in the new tax, since it aims to achieve revenues that can be fully earned abroad, without any financial transit through Brazil. At this point, in addition to the questionable tax jurisdiction of Brazil to tax these revenues, how can the effectiveness of their collection be ensured, considering that digital services taxes are unilateral measures by market countries that are totally disapproved by the countries of residence of the potential taxpayers (mainly the US)?

Obviously, there would be no collaboration between these countries’ and the Brazilian authorities, which would make collection practically impossible in many cases. This problem would also affect equality, since multinationals formally established in Brazil could not compete on equal terms with companies that operate exclusively from abroad, which could lead to a flight of these multinationals from Brazil (and, with that, loss of jobs and the reduction of taxes revenues, which, once again, undermines the national development principle).


The reasons explained above are more than enough for the CIDE-Digital tax not to be enacted into law. It should be remembered that the import of services, including those considered “digital,” are already subject to a huge tax burden in Brazil, considering the income tax withheld at source, CIDE-Technology, PIS/Cofins, Services Tax (ISS) and sometimes the state VAT (ICMS). Therefore, the peculiar Brazilian situation is totally different from that of the countries that implemented the DST, since these nations do not tax international remittances in the way the Brazilian government does.

In addition, whether due to the heavy tax burden of international remittances or other peculiarities of Brazil (high level of unbanked people, low percentage of consumers with international credit cards, seasonal instability of the exchange, etc.), it is a fact that the largest multinational technology companies are formally established in Brazil as Brazilian legal entities and, therefore, are already subject to the country’s heavy tax burden. This point also puts Brazil in a different position from that of the other countries that have already instituted their DSTs, which means that the proposed Brazilian DST is of little practical use except to be another source of government revenue.

For all these reasons, Draft Bill 2358/2020 must be rejected by the National Congress.

Worst of both worlds: A case against digital services tax in Brazil

Maurício Barros is a Partner at Gaia Silva Gaede Advogados in São Paulo. Barros is also a  former Taxpayer-Appointed Judge at the São Paulo Taxes and Fees Court – TIT/SP (2014-2019) and a former Visiting Professor at the Getulio Vargas Foundation and at the Mackenzie Presbiteryan University.


Luiz Guilherme de Medeiros Ferreira is a tax lawyer in São Paulo and Member of the Tax Litigation Commission at the Brazilian Bar Association.