BNLData – FactCheck: How did the IEPS study transform an assumption into R$ 38,8 billion in "losses" for betting companies?
The sports betting and online gaming sector has been the target of criticism based on dubious and suspicious studies, which end up becoming narratives and are cited in various reports, articles, and speeches by politicians, especially when they have a negative bias.
Currently, one of the most cited figures, and one that has dominated the debate on gambling regulation in Brazil in recent weeks, is R$ 38,8 billion . This is the amount that the Institute for Studies on Health Policies (IEPS), in partnership with the Mixed Parliamentary Front for the Promotion of Mental Health (FPSM) and Umane, attributed to the "social cost" of betting in Brazil—more than triple what the sector collects in taxes. This figure became headline news in dozens of media outlets, was cited in congressional hearings, and began circulating as definitive proof that the country pays too high a price for the legalization of online betting.
BNLData – FactCheck examined the original documents of the study and found a problem that none of the news reports that reported on the number mentioned: the authors of the report themselves acknowledge, in a technical note attached to the graph that supports the figure, that it is not possible to prove that the losses recorded were caused by the bets.
The phrase that nobody repeated.
The report “The Health of Brazilians at Stake: A Political-Economic Analysis of the Regulation of Online Betting and its Impacts on the Health of the Brazilian Population,” released on December 2, 2025, is the source of all the figures. It details that R$ 30,6 billion of the R$ 38,8 billion comes from health damages—with R$ 17 billion attributed to suicide deaths and R$ 13,4 billion to depression. It is precisely alongside these numbers, in the footnote of the official report's graph, that the caveat appears that would change the interpretation of all subsequent coverage:
“People with problematic gambling use suffer a higher incidence of depression and suicide, among other events. The estimates represent the costs related to cases of suicide and depression associated with gambling. They are associated with gambling because, given the limited evidence, it is not possible to state that such harm was directly caused by problematic gambling.” — Technical note from the IEPS/FPSM/Umane dossier
In the language of the research itself, "associated" and "caused" are not synonyms—and the difference between the two concepts is what separates a hypothesis from a fact. Even so, the final result of the calculation was presented, and widely replicated, as if it were a real and measurable cost that betting imposes on Brazilian society.
A formula imported from England
If the data doesn't come from Brazilian cases proven to be caused by gambling, where does it come from? The answer lies in the methodology described by IEPS itself: the cost values per gambler profile — low risk, moderate risk, and problem gambling — were taken from a 2023 study by the Office for Health Improvement and Disparities (OHID), a British government agency.
These values in pounds were converted to reais using a purchasing power parity rate (R$ 3,40/£) and then multiplied by the proportion of Brazilians at risk from gambling, as measured by the third National Survey on Drug Use (LENAD III) — a survey on substance use, not a study designed to measure the causal impact of gambling on mental health. The result is a three-layered extrapolation: British data, converted using the exchange rate, applied to a Brazilian proportion obtained by analogy.
There is no point along these lines where Brazil has directly measured how many suicides or depression diagnoses result from gambling. The number is a projection—legitimate as a technical exercise, but treated by the media as a field survey.
QALY: Statistics dressed as a fait accompli
The R$10,4 billion attributed to "loss of quality of life due to depression" uses the concept of QALY (quality-adjusted life years), a standard tool in health economics for comparing the impact of different diseases. The problem is not the tool itself, but the way the result was communicated: a lost QALY becomes real money based on assumptions—monetary value assigned to a year of life, discount rate, life expectancy—that the disclosure material does not detail and which, if altered, change the final result by billions. Presenting this value as a "cost" alongside direct expenses such as medical treatment (R$3 billion, which is a real outlay) conflates two very different types of data under a single label.
The comparison that doesn't add up.
The most frequently repeated contrast in the coverage — "the social cost is triple the revenue" — compares a statistical estimate (R$ 38,8 billion) with a real and auditable fiscal value (R$ 6,8 billion collected between January and September 2025, according to the Federal Revenue Service). These are magnitudes of different natures: one is measured, the other is projected. Placing them side by side without this caveat creates the impression of a "number that doesn't add up," which is, in fact, a comparison between an accounting certainty and a statistical hypothesis.
What withstands scrutiny
Not everything in the report is flawed. It is verifiable and correct that only 1% of tax revenue from betting is earmarked for healthcare—compared to 50% in the United Kingdom, according to the study itself. It is also a fact that problem gambling exists and demands public policies for prevention and treatment, regardless of any calculation of social cost. The error lies not in pointing out the existence of the problem, but in cloaking it in numerical precision that the research itself does not support.
Conclusion
The figure of R$ 38,8 billion is the result of a three-step statistical extrapolation, built upon a relationship that the authors themselves classify as an association, not a cause. Treating it as a fait accompli—as almost all coverage of the study did—is not a typo or an isolated editorial exaggeration: it is the transformation of a hypothesis into political ammunition, without the uncertainty acknowledged by the researchers themselves reaching the public.
