Computer & Communication Industry Association
PublishedNovember 17, 2026

Brazil’s R$ 50 Billion Promise: A DMA Bill Benefit Estimate that Fails its Own Stress Test

Brazil’s Ministry of Finance has handed lawmakers a technical note with a dubious headline number for its DMA-inspired digital markets bill, PL 4675/2025: R$ 15.3 billion to R$ 50.6 billion a year in claimed economic benefits, or as much as R$ 240 per inhabitant. The headline number does not survive a careful reading of the Ministry’s own paperwork. If you rerun the benefits calculation with the alternative inputs printed in the Ministry’s own methodological annex, the range collapses to roughly R$ 5 billion to R$ 15 billion, before even counting the costs. The technical note estimate multiplies inflated revenue by a price cut in a manner copied from cartel cases, then books a transfer as if it is a net gain, and fails to quantify any costs. Congress should not legislate under an urgency procedure on such fragile arithmetic.

What methodology does the Brazilian Ministry of Finance use to estimate the benefit from the bill? A questionable one. The technical note takes the estimated 2025 revenue of 13 digital segments, R$ 153.3 billion, and multiplies it by a price cut of 10% to 33% that the note itself describes as illustrative. That is the whole model. It works only if a tenth to a third of every real collected is an overcharge that regulation can hand back without any impacts on service quality or the range of services offered. The note never justifies this assumption. Its own fine print concedes that the note does not independently demonstrate market power in these segments, and that the revenue included in the estimate may not match the revenue the bill would actually touch. 

The note also describes its result as a gross gain to users through a transfer of surplus from platforms. Economists generally do not score transfers as net benefits, because what one side gains, the other loses. Any net gain in total economic surplus would come from efficiency effects that the note says it did not estimate. The press release emphasized a big headline number and ignored the relevant caveats that make the number questionable at best. For good measure, the note touts platform revenue as 3.2% of GDP, a comparison of revenue with value added that its own footnote admits is not a measure of economic contribution to GDP.

The technical note’s 33% ceiling comes from the mean overcharge in a survey of price-fixing cartels, while the note insists that it accuses no platform of unlawful conduct. That mean is pulled up significantly by outliers. Swap in the same survey’s median instead of the mean, and the Ministry’s own annex trims the top of the range to R$ 28.7 billion. A peer-reviewed reanalysis that corrects earlier cartel data for bias puts the typical overcharge near 16%, about a third of the 49% overcharge figure the Ministry used to estimate its ceiling. 

That’s not all: The 10% floor is actually above the note’s own merger benchmark. The merger study it cites finds a mean price increase of about 7%, which the annex converts to a 6.6% cut and a benefit of R$ 10.2 billion. The Ministry published 10% regardless, relying instead on nominal app-store commission cuts. 

Finally, the headline contradicts the note’s own Table 3, which reserves cuts above 25% for near-monopoly markets that have already tipped and caps markets that are still forming at 20%. The headline applies 33% to every selected digital segment, artificial intelligence included, despite the enormous dynamism and entry found in the digital economy generally and in the artificial intelligence sector particularly.

The only platform-specific price effect evidence provided by Brazil’s Ministry of Finance is three app store episodes, in South Korea, the Netherlands, and the European Union, where the headline commission on some payments fell from 30% to between 20% and 27%. The annex admits these are nominal cuts that ignore what developers pay their new payment processor and how many developers actually switched. In other words, the technical note treats a nominal price change, before accounting for offsetting incremental costs or actual adoption rates, as support for a benefit estimate, and then also fails to quantify any offsetting costs.

Even at face value, a commission cut tells us nothing about services that charge no commission. App stores are under 5% of the Ministry’s revenue base. The rest is mostly cloud subscriptions, advertising, software licenses, and ride and delivery fees. Advertising alone is nearly a quarter of the base, and it funds services that users receive for free. The note concedes in a footnote that its result holds only if a price cut on one side of a platform is not offset on another. 

The annex quietly lays the groundwork for any reasonable reader to conclude that the revenue base is padded. Cloud, the largest line item at R$ 66.3 billion, bundles subscription software with infrastructure; strip out the software and cloud shrinks to between R$ 15.1 billion and R$ 27.4 billion. The ride-hailing figure may be total fares, in which case platform revenue would be closer to R$ 3.7 billion than R$ 14.95 billion. PC operating systems could be R$ 1.4 billion instead of R$ 11.9 billion. The mobile operating system line turns out to be an assumed 10% of iPhone sales, despite being a price that no one directly pays. 

Apply those corrections together and, in the Ministry’s own table, the base falls from R$ 153.3 billion to R$ 80.4 billion. Pair that base with the Ministry’s own alternative price parameters, 6.6% and 18.7%, and the range becomes R$ 5.3 billion to R$ 15.0 billion, which is only about a third of the headline number.

The technical note prices benefits to the decimal and does not quantify costs at all. The annex supplies a single figure, borrowed from a study of direct compliance costs, and then acknowledges that dynamic costs such as delayed product launches, which have become a major issue in Europe after the enactment of the original EU DMA, are not quantified. They can be, and they should be, quantified. That same study estimates the consumer welfare loss from a one-year delay in new products at R$ 8.7 billion in present value, summed over ten years, a cost that appears nowhere in the Ministry’s arithmetic. Europe’s experience shows the cost is real: Apple has postponed features including iPhone Mirroring and Live Translation for EU users, citing the Digital Markets Act.

The bill does not directly set prices, but it empowers the competition authority, CADE, to impose obligations on designated firms on a case-by-case basis. The technical note applies the assumed price cut to the entire selected revenue base, without quantifying the share that would actually be affected, the implementation timetable, or the likelihood of achieving the assumed reductions. It also fails to separate the bill’s incremental effects from enforcement under existing law and ongoing market competition. CADE opened an inquiry into cloud licensing practices in January 2026 under current law. In food delivery, two well-funded rivals arrived in 2025, and the incumbent has since announced a R$ 24 billion investment plan.

Even the Ministry’s chosen yardstick cuts against it. The annex notes that the European Commission projected consumer gains of about 0.09% of GDP for the Digital Markets Act. Brazil’s headline claims 0.12% to 0.40%, more than four times at the top end what Brussels promised for the original.

Before voting, lawmakers should ask the Ministry for four things: 

  • evidence of market power and overcharges segment by segment; 
  • price effects calibrated on the business models actually being regulated; 
  • a full accounting of costs, including the ones consumers would feel in delayed, degraded, or withdrawn services; and
  • estimated impacts on investment in Brazil over the medium-term. 

Brazilian consumers deserve low prices, good services, and policy built on evidence. An estimate that loses about two-thirds of its value under the Ministry’s own alternative assumptions and fails to quantify enormous potential costs imposed by the bill on users is no basis for rewriting Brazil’s competition law.

Trevor Wagener

Director of the Research Center & Chief Economist, CCIA
Trevor Wagener is the Director of the Research Center & Chief Economist for the Computer & Communications Industry Association, where he leads CCIA’s research agenda, conducts and oversees economic and policy research, and educates policy makers and the public about relevant empirical findings.
Article

A promessa de R$ 50 bilhões do Brasil: uma estimativa de benefícios do PL inspirado no DMA que não passa no próprio teste de estresse

Nota técnica duvidosa alega até R$ 50 bilhões em benefícios de projeto de lei nos moldes do DMA O Ministério da Fazenda entregou aos parlamentares uma nota técnica com um número de destaque ...
  • Competition
Article
AI and Copyrights: Why Using Copyrighted Material for Training is Fair Use

AI and Copyrights: Why Using Copyrighted Material for Training is Fair Use

A collected series of cases captioned as In re OpenAI Copyright Infringement Litigation presents the latest and most important opportunity to resolve an emerging and critical question in copyright law...
  • Copyright
Article
Ending H-1B Grace Period Fails Cost-Benefit Test

Ending H-1B Grace Period Fails Cost-Benefit Test

Executive Summary Costs of ending exceed savings by about $121 million to $762 million a year across the three scenarios, with a central estimate of $263 million. Over ten years, the net cost i...
  • Immigration
Article
Applying Semiconductor and Server Tariffs to Data Centers, Under Section 232 or Section 301, Would Now Cost the U.S. About $180 Billion a Year

Applying Semiconductor and Server Tariffs to Data Centers, Under Section 232 or Section 301, Would Now Cost the U.S. About $180 Billion a Year

An update to CCIA’s June 2026 analysis: the Section 232 data center exemption remains under review, a Section 301 excess capacity determination could be weeks away, and the buildout at stake has gro...
  • Trade