Where does the money go when a country bans betting?
On October 6, 2026 Brazil's 188 licensed betting sites went dark under a presidential decree, and the figure everyone quotes for what households get back is R$30 billion a month. We tested that figure, and what happened to the money after eight earlier bans, from Norway to India.
Where does the money go when a country bans betting?
On September 25, 2026 President Lula signed Medida Provisória 1.394, a decree with the force of law that bans fixed-odds betting, online casino games and every advertisement for them (Planalto, 2026). Deposits stopped that day, the 188 licensed sites went offline on October 6, and the licences themselves end on October 25 (Justice Ministry, 2026; SPA, 2026c). The President called betting a cancer that takes the money for milk and bread (Agência Brasil, 2026a). The retail federation CNC, whose April study is the source of the figure everyone repeats, said Brazilians were spending more than R$30 billion a month on betting sites and that shops had lost R$143 billion in sales since 2023 (CNC, 2026). The claim tested here is the one the decree, the retailers and the sell side share: that the ban puts that money back into households, and from there into shops.
The ruler is the gap between what passes through a betting site and what stays there. A bettor who deposits R$100, wins R$150 and bets it all again has staked R$250, deposited R$100 and, if the last bet loses, lost R$100. Stakes, deposits and Pix transfers count the passing-through; the household’s loss is the operators’ gross gaming revenue, stakes minus prizes, which the Finance Ministry publishes. In the first half of 2026 the licensed operators kept R$20.1 billion from R$88.5 billion of deposits, about R$3.3 billion a month (SPA, 2026a; Folha de S.Paulo, 2026).
The claim fails on its number, on the record of earlier bans and on where the money comes from. Brazilian households lose about R$3.3 billion a month to the licensed operators, and about R$4.7 billion by a count of Pix flows in both directions: a sixth to a ninth of the R$30 billion, and half a percent of what they spend. No earlier ban cut a country’s gambling loss by half, the bar set in advance. Norway, which removed every slot machine in 2007, saw losses fall 28% in a year and stay 17% lower five years on; Britain cut the stake on its betting-shop machines and 57% of the lost spend moved online. Studies of the opposite experiment, the opening of betting in 26 US states, found the money came out of savings and credit, with other spending unchanged (Baker et al., 2024). What a ban returns goes to debt and savings before it reaches a shop.
The R$30 billion a month is what passes through betting sites; what stays is R$3 to 5 billion.
Every public figure for Brazil's betting money, on one monthly scale, R$ billion: stakes and deposits at licensed operators, Pix transfers as the Banco Central and Comsefaz measured them, the CNC's figure, and the two measures of the households' loss (accented bars).
How we tested it
The claim, its tests and the rules for the verdict were written down and fixed on October 7, 2026, before any data for them was pulled. The claim has three parts, and each was given a bar to clear (Fig. 2).
The money comes first. Every public figure for Brazil’s betting flows goes on one monthly scale, with the gross ones (stakes, deposits, Pix transfers) apart from the households’ net loss (the operators’ gross gaming revenue, or the net Pix transfer where a study measured both directions). The part holds if the loss is at least a third of the quoted figure, and fails if the figure is three times the loss or more. The bar of three was fixed with the first-half revenue figure already in hand, as the tail records.
Earlier bans come second. For nine cases named in advance, from Norway’s removal of every slot machine in 2007 to India’s ban on real-money gaming apps, the test takes the affected product’s spend in the 12 months after the measure against the 12 months before, the spend in the nearest substitutes over the same window, and the country’s total gambling loss, as stakes minus prizes where the regulator publishes it. The part holds if, in the median case, total gambling fell by more than half and stayed down; it fails if substitutes absorbed more than half of what the banned product lost. The Philippines drops out: its offshore operators served customers in China, so their closure moved no Filipino household’s money. Figure 8 names the four cases whose series break the 12-month window. The reading that divides the cases into products that went and products that stayed reachable was made after the results were in, and is marked [post-hoc] where it appears.
Where the money comes from is the third part. The published studies of the reverse experiment, the state-by-state opening of online betting in the United States, give what happened to consumption, saving, investment and debt, and Brazil’s household balance sheet sits beside them. The part holds if the studies find betting was financed out of consumption, so that a ban returns it there, and fails if it was financed out of saving and credit.
The size part fails by a factor of six to nine, the earlier-bans part clears neither of its bars, and the financing part fails.
The three parts of the claim, the bar each had to clear, fixed in advance, and the result.
| Part of the claim | Holds if | Result |
|---|---|---|
| How big the money is | The households' loss is at least a third of the quoted R$30 billion a month | Fails: licensed operators keep R$3.3 billion a month and the Pix count puts the net loss at R$4.7 billion; every gross measure runs 4.4 to 6.7 times its own loss figure |
| Where the money went after earlier bans | Total gambling loss fell by more than half in the typical case and stayed down; fails if substitutes took more than half of the lost spend | Neither bar met: the median fall was 31%, at most 46% in any case, and substitutes took 31% in the median case, 57% in Britain |
| Debt first or shops first | Studies of legalisation find betting was financed out of consumption | Fails: US households cut investment $53 to $81 a quarter and ran up card debt, with other spending unchanged; Brazilian bettors' arrears rose 39% |
Two supporting tests run on proxies. The Banco Central publishes Pix flows by the receiving company’s economic section, not by betting, so the post-ban Pix read is the series for the arts, sport and recreation section, which the licensed operators joined in January 2025 (BCB, 2026b). The balance of payments has no gambling line, and since 2025 the operators have been Brazilian companies, so the flow abroad does not show there either (BCB, 2026a). The read of shops after a measure, retail sales in Britain and household consumption in Norway, is in the robustness section.
The R$30 billion is what passes through; R$3 billion is what stays
Brazil’s betting money is four years old. Fixed-odds betting was legalised in December 2018 and left unregulated until a December 2023 law set a 12% levy on operators’ revenue and a licensing regime that took effect on January 1, 2025; until then the operators were foreign companies paid by Pix, and the Banco Central’s September 2024 study had to identify them by their transfer patterns (BCB, 2024; Planalto, 2026). That study is where the public numbers began. It found Pix transfers of R$18 to 21 billion a month to 56 betting companies, and said in the same paragraph that about 15% of what is bet stays with the operators, so that August 2024’s R$20.8 billion of transfers was about R$3.1 billion of loss (BCB, 2024). From January 2025 the licensed operators became Brazilian companies in the Banco Central’s arts, sport and recreation section, and Pix payments to that section jumped from R$4.5 billion in August 2024 to R$28.9 billion in January 2025 and ran at R$27 to 39 billion a month through July 2026 (Fig. 3).
Pix payments to the sector that holds the betting operators ran at R$27 to 39 billion a month through 2025 and 2026.
Pix payments from people to companies in the arts, sport and recreation section, R$ billion a month, January 2024 to July 2026; the dashed verticals mark January 2025, when licensing made the operators Brazilian companies in this section, and September 25, 2026, the decree.
Three numbers describe the licensed market, and they differ by a factor of twenty. In the first half of 2026 bettors staked R$411 billion at the licensed operators, deposited R$88.5 billion, and lost R$20.1 billion (SPA, 2026a; Folha de S.Paulo, 2026). Stakes count every re-bet of winnings and run at twenty times the loss; deposits count money that comes back as withdrawals and run at four to six times it; only the last number, the gross gaming revenue, is money that left households.
The figures in public debate are of the first two kinds, and they count the unlicensed sites as well. The state finance secretaries’ council, Comsefaz, measured both directions of the Pix flow to the sector and found R$24.1 billion a month going in and R$4.7 billion a month staying, five times less (Comsefaz, 2026). The CNC’s R$30 billion is a gross figure of the same family; its study does not say which (CNC, 2026). The honest comparison is each gross measure against its own loss: the Banco Central’s transfers run 6.7 times the loss it implies, the Comsefaz flow 5.1 times, and the licensed deposits 4.4 times. Against the licensed loss the CNC figure is nine times too high, and against the Comsefaz loss six times. If the unlicensed sites, which the industry’s own survey put at 38% to 44% of play, lose money at the same rate as the licensed ones, households lose R$6 to 7 billion a month to betting in all, a fifth of the figure (IBJR, 2026).
The loss has grown, and it is small against what households spend. The licensed operators kept R$37.0 billion in 2025 and R$20.1 billion in the first half of 2026, 15% more than a year earlier (SPA, 2026b; SPA, 2026a). Brazilian households spend about R$694 billion a month, so the licensed loss is 0.48% of consumption and the Comsefaz loss 0.68%; the CNC’s figure would be 4.3%, and the stakes figure a tenth of everything households buy. Before licensed betting existed, the 2017 household budget survey put all gambling, mostly lotteries, at 0.27% of consumption (TCU, 2025). The state’s share is smaller still: the gambling division paid R$10.0 billion of federal tax in 2025, and the government expects the ban to cost it R$1.5 billion in 2026 and R$5.2 billion in 2027 (Receita Federal, 2026; Planalto, 2026b).
The money is also smaller than the budget it is most often set against. In August 2024 five million people in Bolsa Família households sent R$3 billion to betting companies, a fifth of the R$14.1 billion the programme paid that month (BCB, 2024; MDS, 2024). At the Banco Central’s own retention rate, the loss behind those transfers was about R$450 million, 3% of the payroll.
Who sends the money matters as much as how much. Forty-four million Brazilians, 27% of adults, had placed at least one bet with a licensed operator by August 2026, though only 8.5 million deposited money in a typical month of the first half (Folha de S.Paulo, 2026). The money concentrates far beyond that, because the typical bettor sends little: in the Banco Central’s August 2024 data the youngest bettors sent about R$100 a month on average and the oldest more than R$3,000, and among Bolsa Família bettors the median was R$100 (BCB, 2024). The audit court’s follow-up in January 2025 found a median of R$100 and a mean of R$628 among the programme’s bettors, with the 19% who sent more than R$600 accounting for 84% of the money (TCU, 2025). In the Firpo and Stone study of people who began betting after January 2025, one bettor in ten sent more than a fifth of everything leaving their account to betting sites (Jornal do Comércio, 2026).
Among Bolsa Família bettors a fifth send four fifths of the money, and across the country a fifth of those who ever bet deposit in a given month.
How many Brazilians bet and how the money concentrates, nine figures with their sources.
| Measure | Figure | Source |
|---|---|---|
| Adults who ever bet, August 2026 | 43.8 million | Finance Ministry data under the access-to-information law (Folha, September 18 2026) |
| Bettors depositing in an average month, H1 2026 | 8.5 million | Finance Ministry data under the access-to-information law (Folha, September 18 2026) |
| Households with a bettor, Datafolha, October 1 2026 | 18% | Datafolha, fieldwork September 28 to October 1 2026 |
| Bolsa Familia households with a bettor, same poll | 31% | Datafolha, fieldwork September 28 to October 1 2026 |
| Median monthly transfer per Bolsa Familia bettor, August 2024 | R$100 | BCB Estudo Especial 119/2024 |
| Mean transfer per Bolsa Familia bettor, January 2025 | R$628 | TCU, TC 024.146/2024-2 |
| Share of Bolsa Familia bettors above R$600 a month | 19.2% | TCU, TC 024.146/2024-2 |
| Their share of Bolsa Familia bettors' money | 84.2% | TCU, TC 024.146/2024-2 |
| Bettors who sent more than 20% of monthly outflows to bets | 1 in 10 | Firpo (Insper) and Stone, as reported September 8 2026 |
That shape decides who feels a ban. For the median bettor, R$100 a month of transfers, most of it returned as prizes, is a few reais of loss and a habit; a ban changes an evening, not a budget. For the tenth who sent a fifth of their outflows, the loss was the budget, and where that money goes is the question of this report. Datafolha’s poll in the week of the ban found a bettor in 18% of households and 78% of respondents in favour of the ban (Datafolha, 2026).
Checking your access.
Keep reading for free.
Check your email.
Opening the study.
Leave your email to unlock the full study.
Your membership has ended.
Renew to read all our research in full again.
Renew