Online sports betting raised US bankruptcy risk 25–30% and cut household investment 14%, yet it supplies 0.2% of state tax revenue. Evidence from six countries.
The Seven-Year Experiment
How one court ruling turned sports betting into a mass-market product
On 14 May 2018 the US Supreme Court struck down the federal law that had confined sports betting to Nevada since 1992 ✓ Established [1]. By December 2025, 39 states and the District of Columbia had legalised it [2], and the amount Americans wagered legally each year had grown from $4.9 billion in 2017 to $166.94 billion in 2025 [3] [4]. The evidence now available shows what that growth cost, who paid for it, and how little of it reached public budgets.
The core question of this report has a short answer. Legalisation did not simply move an existing illegal market into the light. It created a new, mostly mobile, highly concentrated consumer product, and the best evidence finds that the product damaged household finances in measurable ways ◈ Strong Evidence. Studies using bank transaction data and credit reports find lower savings, more overdrafts, more debt collections and a 25% to 30% rise in bankruptcies in states that allowed online betting [5] [6]. Searches for gambling-addiction help rose 23% nationally after the 2018 ruling [3]. The harm is concentrated among young men and low-income households [6]. Against this, the tax revenue is real but small: sports betting supplies about 0.2% of average state tax revenue [7]. Britain and Australia, which liberalised gambling much earlier, show what a mature version of this market looks like — heavy advertising, revenue drawn from a small share of heavy losers, and governments that have repeatedly stepped back from the strongest reforms their own inquiries recommended [8] [9].
The scale of the American market is now comparable to large established industries. In 2024, US sportsbooks earned $13.71 billion in revenue on $149.90 billion of legal wagers, and more than 95% of the money was bet online [10]. In 2025 revenue rose 22.8% to $16.96 billion on a record $166.94 billion in wagers, and state taxes on sports betting rose 32.4% to $3.71 billion [4]. Sports betting is now one part of a commercial gaming industry that took $78.7 billion in revenue in 2025 [11]. The distinction between the amount wagered, called the handle, and the amount the operator keeps, called revenue or gross gaming revenue, matters throughout this report. Handle measures activity. Revenue measures what customers lost. In 2025 revenue grew twice as fast as handle, which means that each dollar bet became more profitable for the operator ✓ Established [12].
The market is also unusually concentrated. In the second quarter of 2025, FanDuel reported $1.79 billion in revenue and DraftKings $1.5 billion, against $692 million for BetMGM, the third-largest operator; together the two leaders took nearly 83% of the adjusted earnings of the six largest operators [13]. Of the 39 legal states, 31 plus the District of Columbia allow betting on a phone, while eight permit only in-person betting at a casino or retail sportsbook [14]. That distinction turns out to be the most important variable in the research. Almost every study that separates the two finds that the financial and health effects are much larger where mobile betting is legal [6] [3] [15]. A betting window in a casino is a product people travel to. An app is a product that travels with them, available at any hour and during the game itself. The policy debate in 2018 was mostly about whether to legalise. The evidence since then suggests that the more consequential choice was how — and in particular whether to allow betting on a phone.
The case for legalisation rested on three promises: that it would absorb an existing illegal market, that it would generate tax revenue for public services, and that regulated operators would protect consumers better than offshore websites and local bookmakers. The American Gaming Association, the industry's trade body, estimated in 2022 that Americans bet $63.8 billion a year with illegal bookmakers and offshore sites ⚖ Contested [16]. Yet legal wagers grew roughly 34-fold between 2017 and 2025 [3] [4]. Even if the industry's estimate of the illegal market is accurate, most legal betting is new activity, not displaced activity. The clearest sign is the household data: one study of 230,000 households found that online betting did not reduce spending on other forms of gambling or consumption, but did reduce savings [5]. People did not switch from a bookmaker to an app. They started betting, and the money came out of what they would otherwise have saved.
Four findings now have support from more than one independent dataset, and they organise the sections that follow:
- Household finance. Bank and credit data show lower savings, more overdrafts, more delinquencies and more bankruptcies after online betting became legal, concentrated in financially constrained households [5] [6] [17].
- Health signals. Help-seeking searches, helpline contacts and mental-health treatment all rose after legalisation, most strongly where mobile betting was allowed [3] [18] [19].
- Advertising saturation. Britain, Australia and the United States each saw a surge of betting advertising around live sport, with young people among the most exposed [20] [21] [22].
- A modest fiscal return. Sports betting tax revenue has grown quickly from a low base but remains about 0.2% of state tax revenue, and some of it is offset by lower lottery sales [7] [23].
The international comparison gives the American story a time dimension. Britain opened television advertising to sports betting and online casinos in September 2007 under the Gambling Act 2005, and gambling advertisements on British television rose from 62,000 in 2005 to 1.39 million in 2012 [20]. Australians lost A$33.0 billion on legal gambling in 2024-25, A$1,528 per adult, the highest per-person losses recorded anywhere [24]. Brazil legalised online fixed-odds betting in December 2023 and opened a regulated market in January 2025 [25]. Japan, by contrast, decided in December 2025 to keep sports betting illegal, with a cross-party group of lawmakers citing match-fixing and athlete protection [26]. Each country offers evidence on one stage of the same process: the rapid growth of a product, the political arrival of its costs, and the difficulty of regulating an industry once it is embedded in the finances of sport and broadcasting.
This report uses three levels of evidence. Claims marked as established rest on official statistics or on large studies that have been replicated or published in peer-reviewed journals. Claims marked as strong evidence rest on well-designed studies that have not yet been replicated, or on consistent findings from several weaker sources. Claims marked as contested are those on which credible sources disagree, and both sides are given. Most of the American research uses the staggered timing of legalisation across states as a natural experiment: it compares changes in states that legalised with changes in states that had not yet done so. This method cannot prove causation in the way a randomised trial can. But when studies using different data — bank transactions, credit reports, insurance claims and internet searches — arrive at the same direction of effect, the burden of proof shifts towards those who argue that legalisation was harmless ◈ Strong Evidence [5] [6] [19] [3].
How a Sportsbook Makes Money
Hold, parlays and the customers who lose the most
In June 2025 US sportsbooks kept a record 12.4% of all money wagered ✓ Established [27]. That figure, called the hold, has risen steadily since 2018 because operators have steered customers towards complex multi-leg bets on which the bookmaker's margin is much larger. The business also depends on a small share of customers: in Britain, the top 10% of online accounts generated 79% of operator revenue [28].
A sportsbook does not, in principle, need to predict the outcome of games. It sets prices — odds — that include a margin, and if it attracts roughly balanced betting on each side, it keeps the margin whatever happens. On a simple bet on one team to beat another, the margin is typically around 4% to 5% of the amount staked. In practice, US operators have become far more profitable than that. In 2025, handle grew 11.0% while revenue grew 22.8%, because the average hold rose to about 10.2% from about 9.2% a year earlier ✓ Established [12]. The record month came in June 2025, when operators across 31 states kept 12.4% of just over $9 billion in wagers [27]. The higher hold did not come from better luck. It came from a change in what customers bet on. Operators have pushed the parlay — a single bet that combines several outcomes, all of which must be correct for the bet to win — and especially the same-game parlay, which combines several outcomes within one match.
The mathematics of the parlay explains the strategy. The margin on each part of the bet is multiplied, not added, so a five-leg parlay carries a far larger house edge than five separate bets. The customer sees a small stake and a very large potential payout. The operator sees a product with a high and stable margin. DraftKings told investors that its parlay mix in June 2025 was 430 basis points higher than a year earlier [27]. Its full-year 2024 results attributed the rise in its sportsbook net revenue margin, from 6.0% to 7.1%, to a higher parlay mix, lower promotional spending and favourable outcomes [29]. In September 2025 the company introduced a feature for NFL parlays, which it called Ghost Leg, designed to encourage still more of them [30]. The direction of product design is clear: towards bets with the highest margin, presented in a form that makes the cost hardest for the customer to see ◈ Strong Evidence.
The Patterns of Play study analysed 139,152 British online gambling accounts and found that the top 10% of accounts generated 79% of operator revenue, and the top 20% just over 90% [28]. In each of betting, slots and casino games, the 5% of accounts with the highest losses produced at least 70% of gross gambling yield [31]. In Massachusetts, 6.1% of bettors accounted for 51.3% of net losses [32], and an industry analysis of 1.28 billion NFL bets found that 3% of players generated 82% of sportsbook revenue [33].
These figures change the meaning of the industry's standard defence, that the great majority of customers bet small amounts for entertainment. That statement is true, and it is also beside the point. If most revenue comes from a small minority of customers, then the commercial interest of the operator is tied to that minority, and the people in it are disproportionately those who struggle to control their betting. In the Patterns of Play survey, 19% of the account holders with the heaviest losses reported difficulty cutting back, compared with 2% of those whose accounts were in profit [28]. The same study found that gaming accounts were nearly three times as common among residents of the most deprived areas of Britain (29%) as among those in the least deprived (13%), and that the most deprived fifth of the population supplied a quarter of gaming losses [28]. A business model that depends on heavy losers will, other things being equal, draw its income from those least able to afford it ◈ Strong Evidence.
The other side of this model is the treatment of customers who win. Every US state allows sportsbooks to limit the stakes of customers they judge to be skilled, and only New Jersey bars them from closing such accounts entirely [34]. Data from the Massachusetts Gaming Commission showed that 0.64% of sports betting accounts in the state had been limited, nearly 58% of them to between 1% and 24% of the normal maximum stake, and that above-average winnings were associated with a higher chance of restriction [35]. In September 2025 the commission voted unanimously to advance a rule requiring operators to tell customers when they have been limited [35]. Reporting by the Wall Street Journal described the reverse treatment for heavy losers: VIP hosts, larger betting limits and personal attention [34]. A sportsbook therefore filters its customer base in both directions — it restricts those who could take money from it and cultivates those who lose the most.
A conventional retailer wants as many satisfied customers as possible. A sportsbook's accounts show something different: it needs a minority who lose heavily and repeatedly, and it limits customers who win. Consumer protection rules written for ordinary products — disclosure, fair terms, the right to complain — do not address this structure, because the harm is not a defect in the product. It is the source of the profit.
Speed and continuity are the second design feature. Online sportsbooks allow in-play betting, in which customers bet on events within a match while it is happening, and cash-out, which lets a customer settle a bet early for a smaller sum. A scoping review of the research found that people with higher problem-gambling scores are more likely to bet in-play, and researchers have argued that these features turn sports betting from an occasional activity into a continuous one, more like a slot machine than a weekend bet on a result [36]. An experimental study published in Psychological Science in 2024 found that people placed larger bets when a cash-out option was available, and cashed out more often when the odds turned against them than when they improved [37]. Each of these features increases the number of decisions a customer makes, and each decision is another opportunity for the operator's margin to apply.
The Lancet Public Health Commission on gambling, which published the most comprehensive review of the global evidence in October 2024, estimated that 8.9% of adults and 16.3% of adolescents who bet on sports are gambling at a level that carries risk of harm ✓ Established [38]. For online casino products the figures are higher still, at 15.8% of adults and 26.4% of adolescents [38]. Sports betting is not the most dangerous gambling product. But it is the one that has been legalised most rapidly, it is increasingly sold on the same apps as online casino games, and its marketing is attached to the sports that young men watch most.
Anyone with a mobile phone now has access to what is essentially a casino in their pocket, 24 hours a day.
— Heather Wardle, co-chair of the Lancet Public Health Commission on gambling, University of Glasgow, October 2024The Household Ledger
What bank accounts and credit reports show after legalisation
The strongest evidence on the effects of legal sports betting comes not from surveys but from financial records. Two independent studies — one using the bank transactions of about 230,000 households, the other the credit reports of about 7 million consumers — find the same pattern: savings fall, debt rises and bankruptcies increase after online betting becomes legal ◈ Strong Evidence [5] [6].
The first study, by Scott Baker of Northwestern University and four co-authors, used anonymised transaction data from about 230,000 US households and compared changes in states that legalised sports betting with changes in states that had not yet done so [5]. It found that online betting did not displace other forms of gambling or other consumer spending. Instead, it reduced savings. Net investment — money put into brokerage and investment accounts minus withdrawals — fell by about 14% among households that started betting, and each dollar bet was associated with a fall of about two dollars in deposits to investment accounts [39]. The average betting household spent about $1,100 a year on sports bets [39]. Among financially constrained households, the effects went further: credit card debt rose, available credit fell and overdrafts became more frequent [5]. The authors wrote that legal online betting “significantly reduces savings, as risky bets crowd out positive expected value investments” [5].
The second study, by the economists Hollenbeck, Larsen and Proserpio, used credit records from March 2016 to June 2023 and is now published in Management Science [6]. It found that average credit scores fell by about 0.3% in states that legalised sports betting, and by about three times as much in states that allowed online betting [6]. The likelihood of bankruptcy rose by 25% to 30% in online-betting states, the amount of debt sent to collection agencies rose by about 8%, and delinquencies on credit cards and car loans increased [6] [40]. The effects were largest among young men in low-income counties [6]. A fall of 0.3% in an average credit score sounds small. But averages include the large majority who do not bet. When the change is concentrated in a minority of households, as the bankruptcy figures suggest, the effect on that minority is much larger.
Using credit records for 2016 to 2023, Hollenbeck, Larsen and Proserpio found that online legalisation increased the likelihood of bankruptcy by 25% to 30%, raised debt-collection amounts by about 8% and lowered average credit scores roughly three times as much as retail-only legalisation [6]. The effects were concentrated among young men in lower-income areas [40]. The finding is consistent with separate bank-transaction evidence of falling savings and rising overdrafts [5].
A third study, published by the Federal Reserve Bank of New York in February 2026, added a geographical dimension. Using credit data, the economists found that betting rose not only in states that legalised it but also across their borders, with the effect fading within about an hour's drive [17]. Delinquency rose particularly among people under 40: credit card delinquency in that group increased by about one percentage point and car-loan delinquency by about half a point after legalisation [41]. The authors concluded that “distress crosses the border in greater proportion than betting” and that states bordering legal markets “bear a disproportionate share of the harm of legalization while collecting none of the revenue” [17]. That finding matters for the fiscal argument examined later in this report: a state's tax receipts measure only part of the cost, and some of that cost is exported to neighbours.
The most recent evidence reaches the most basic household need. A working paper published by the National Bureau of Economic Research in June 2026 studied the effect of legalisation between 2021 and 2023 on food sufficiency — whether households have enough to eat [42]. It found that food sufficiency fell by 2.1% among working-age adults without a university degree, and by 10.5% among active bettors [42]. The share of people who said they found it hard to pay their usual household expenses rose by 3.7 percentage points. The authors estimated 284,000 additional food-insufficient households and $130.2 million in additional health spending each year across the nine states they studied, and they identified financial distress, not mental health, as the main channel [42]. The paper has not yet been peer-reviewed, and its figures should be treated as strong rather than established evidence.
Financial strain does not stay within household accounts. Researchers at the University of Oregon used police crime data for 2011 to 2022 to study what happens after an unexpected loss by a local NFL team. Such upsets were already known to be followed by a rise in intimate partner violence. The study found that legalisation increased that effect by about 9 to 10 percentage points, and by more where mobile betting was legal ◈ Strong Evidence [15]. A separate study in the Journal of Sports Economics found more assaults and thefts on game days after legalisation, with spillovers into neighbouring counties [43]. These studies are recent, and the mechanisms they propose — money lost on a game adding to the anger of a loss — need further confirmation. But they are consistent with the financial evidence: the costs of a betting loss fall on people close to the bettor, not only on the bettor.
The household evidence can be summarised in five findings, each from a different dataset:
- Savings. Net investment fell about 14% among households that began betting [39].
- Credit. Credit scores fell and debt collections rose about 8% after online legalisation [6].
- Bankruptcy. Bankruptcy likelihood rose 25% to 30% in online-betting states [6].
- Delinquency across borders. Delinquency rose among under-40s, including in neighbouring states that collect no tax [17] [41].
- Food. Food sufficiency fell 10.5% among active bettors [42].
None of these studies is a randomised trial, and each could in principle be affected by other changes that happened at the same time as legalisation, such as the pandemic, which coincided with the launch of betting in many states. The authors address this by comparing states that legalised at different times and by testing whether outcomes were already diverging before legalisation; in the main results, they were not [6] [5]. The more important point is that the studies were carried out by different teams, using different data and different methods, and all of them find that legal online betting worsened the finances of a meaningful minority of households ◈ Strong Evidence. No study of comparable quality has found the opposite. The industry's argument that legal betting is a form of entertainment that customers can afford is true for most of its customers. The financial records show that it is not true for all of them, and that those for whom it is not true are the ones who supply most of the revenue [28].
Young Men and the Harm Signal
Helplines, help-seeking and the age profile of risk
Searches for help with gambling addiction rose 23% across the United States after 2018, an excess of about 6.5 to 7.3 million searches ✓ Established [3]. Calls to helplines rose, callers became younger, and in 2025 nearly three quarters of people contacting the national helpline cited financial stress [18]. The group most exposed, in every country examined here, is young men.
Direct measures of problem gambling are expensive to collect and slow to arrive, so researchers have looked for indirect signals that change quickly. The most robust is help-seeking. A study published in JAMA Internal Medicine in February 2025 by researchers at the University of California San Diego found that internet searches for help with gambling addiction rose 23% above the expected level nationally after the 2018 ruling, with about 6.5 to 7.3 million excess searches [3]. In Pennsylvania, searches rose 33% after retail sportsbooks opened and 61% after online sportsbooks opened; in Ohio they rose 67% [44]. The study's authors noted that the association was stronger for online than for retail betting [3]. One co-author, Kevin Yang, said that without systematic monitoring of gambling harm, “we are flying blind while millions bet on sports” [44].
Helpline data point the same way. Calls to the Ohio problem gambling helpline rose 55% in 2023, the state's first year of legal sports betting, and call volume in Pennsylvania more than doubled between 2020 and 2023 [45]. Florida's helpline received 34,000 calls in 2023 and 61,000 in 2024, an 88% increase, and helpline operators reported that callers were increasingly men in their twenties and thirties [46]. Nationally, more than 31,000 people a month contacted the National Problem Gambling Helpline in 2025, contacts had risen 66% over five years, and 73% of those who contacted it cited financial stress, up from 66% a year earlier [18]. Helpline data have limits: more calls can reflect more awareness of the service as well as more harm. But the increase coincides with legalisation in each state, and the growing share of callers citing money problems is consistent with the financial evidence in the previous section ◈ Strong Evidence.
Survey data confirm that the product reaches young men above all. A Pew Research Center survey of 9,916 US adults in July and August 2025 found that 22% had bet on sports in the previous year, up from 19% in 2022, and that 36% of men under 30 had done so [47]. A survey commissioned by the NCAA found that 58% of 18- to 22-year-olds had taken part in at least one betting activity, that 41% of student bettors had bet on their own school's teams and that 35% had used a student bookmaker [48]. In a 2025 survey by the National Council on Problem Gambling, 15% of adults aged 18 to 34 reported problematic gambling behaviour, compared with 2% of those over 55, and 17% of sports bettors did so [49]. Opinion among young men has also shifted. In 2022, 22% of men under 30 told Pew that legal sports betting was bad for society. By 2025 the figure was 47% [47].
The groups with the largest measured harm — young men, low-income households, people already in debt — are the groups that sports betting marketing reaches most directly through live sport, social media and promotions. The overlap is not proof of intent. It does mean that the burden of legalisation falls on people with the least financial margin to absorb losses, and that a policy of individual responsibility asks the most from those least able to provide it.
Clinical data now show the same pattern. A study published in JAMA Psychiatry in October 2026 used about 62 million enrollee-years of employer insurance claims from 2016 to 2023 and found that online sports betting legalisation was associated with higher use of antidepressants and anxiety medication and more outpatient mental-health visits, with larger antidepressant effects among men [19]. It found no change in treatment for substance use or in emergency and inpatient care [19]. Earlier research had established that living in a state with legal betting raised the odds that a person bet on sports by 2.4 to 2.9 times, and that sports betting was associated with moderate or high scores on standard problem-gambling scales [50]. A national survey of 4,363 adults found that sports bettors were far more likely than non-bettors to binge drink at least monthly, by a factor of about nine among men [51].
Britain's official Gambling Survey for Great Britain found that 10.2% of past-year gamblers aged 18 to 24 scored 8 or more on the Problem Gambling Severity Index in 2024, and 10.4% in 2025 [52] [53]. Among all past-year gamblers, the rate for men was 6.0%, compared with 2.8% for women [52]. Across the adult population, 2.4% — about 1.3 million people — scored 8 or more in 2025 [53].
Britain and Australia show what the age profile looks like in older markets. The British survey figures above use an online self-completion method introduced in 2023, and the Gambling Commission's own experiments found that this method produces higher estimates than face-to-face or telephone interviews, so they cannot be compared directly with older surveys that found rates of 0.3% to 0.4% ⚖ Contested [54]. The age gradient within the survey, however, does not depend on that comparison. In Australia, analysis of the long-running HILDA household survey found that the share of men who bet on sports rose from 5.9% in 2015 to 9.3% in 2022, and by more than 60% among men aged 18 to 44 [55]. A 2025 study by the Australian Institute of Family Studies found that more than 70% of regular online sports bettors reported some form of harm, and that only about half used any of the consumer-protection tools offered by operators [56].
The most serious harm is the hardest to measure. A Swedish study of all 2,099 people diagnosed with gambling disorder in national health registers found that their suicide rate was 15.1 times that of the general population aged 20 to 74, and 19.3 times among those aged 20 to 49 [57]. In England, the Office for Health Improvement and Disparities estimated in 2023 that between 117 and 496 suicides a year were associated with gambling, and put the annual economic cost of gambling harm at £1.05 billion or £1.77 billion depending on the method [58]. Those estimates are contested: the statistics regulator asked the department in 2024 to be clearer about the strength of the evidence behind them ⚖ Contested [59]. Globally, the Lancet Commission estimated that about 80 million adults have a gambling disorder or problem gambling, and that 448.7 million gamble at some level of risk [60].
Two further findings explain why the age profile matters. The first is cultural. A 2026 survey by the insurer Northwestern Mutual found that nearly a third of Americans aged 18 to 29 were putting money into, or considering, sports betting and event-contract markets, often alongside cryptocurrency, and that most of them believed riskier routes would get them to their financial goals faster [61]. Researchers have described this as financial nihilism: a belief among young adults that conventional saving will not deliver security, so high-risk bets are rational. The second is institutional. US states spend an average of $0.54 per person a year on problem gambling services, and only seven spend more than $1 [62]. In the 2025 financial year, states allocated about $188 million to such services and spent about $145 million, leaving $43 million unused [63]. The product reached young men within months of legalisation. The treatment system has not caught up.
Saturation
How betting advertising colonised live sport
Over the opening weekend of the 2025 Premier League season, researchers counted 27,440 gambling messages in about 29 hours of broadcast coverage ✓ Established [22]. In Australia, more than a million gambling advertisements aired on free-to-air television and metropolitan radio in a single year [21]. Advertising is the channel through which a legal product becomes a normal part of watching sport, and every country examined here has struggled to restrict it.
Britain provides the longest record. Before September 2007, only bingo, the National Lottery and the football pools could advertise on British television. The Gambling Act 2005, which came into force that month, opened the medium to sports betting, online casinos and poker [20]. The regulator Ofcom later counted 62,000 gambling advertisements on British television in 2005 and 1.39 million in 2012, a more than twentyfold increase in seven years [20]. By 2012 gambling advertising accounted for 4.1% of all the advertising seen by British television viewers [20]. The pattern established then — rapid liberalisation, a surge of advertising around sport, and a long political argument about restricting it — has repeated itself in Australia, the United States and Brazil. In each case the advertising arrived quickly, because operators competing for new customers in a newly opened market spend heavily to acquire them, and the restrictions came slowly, because by then sport and broadcasting had come to depend on the money.
Counting the messages shows what modern betting promotion looks like. Researchers at the University of Bristol counted 6,966 gambling messages during the televised matches of the Premier League's opening weekend in 2022, and 23,690 a year later [64]. In 2025 the count reached 27,440, of which 13,262 appeared during the time the ball was in play, and 2,412 came from 13 brands that were not licensed in Britain [22]. A single match, Wolverhampton Wanderers against Manchester City, carried 5,262 gambling messages — 22 a minute [22]. Most of these were not advertisements in the traditional sense but logos on shirts, perimeter boards and stadium screens, which voluntary codes on broadcast advertising do not cover. The lead researcher, Raffaello Rossi, said that “year on year, the problem seems to get worse, despite the industry's promises of better self-regulation” [22].
Australia's data come from its broadcasting regulator. Between May 2022 and April 2023, more than one million gambling advertisements aired on free-to-air television, regional and metropolitan, and on metropolitan radio, about half of them from online gambling providers; gambling advertisers spent A$238.63 million across these channels and online, 68% of it on free-to-air television [21]. A survey by the Australian Institute of Family Studies found that 78% of adults saw or heard wagering advertisements at least once a week, that advertising had prompted 21% of bettors to start betting and 34% to bet more, and that half of those already at risk had increased their betting in response to advertisements [65]. The parliamentary inquiry chaired by Peta Murphy, which reported in June 2023, concluded that a partial ban on advertising would not work, and recommended a comprehensive ban phased in over three years [8] [66].
Gambling advertising is grooming children and young people to gamble and encourages riskier behaviour. The torrent of advertising is inescapable.
— Peta Murphy, chair, House Standing Committee on Social Policy and Legal Affairs, Parliament of Australia, June 2023The United States compressed the British experience into a few years. Sportsbook operators spent $282 million on national television advertising between September 2021 and May 2022 alone, as New York and other large states opened online markets [45]. National television spending has since levelled off and begun to decline, from about $477 million in 2024 to $442.5 million in 2025, although total advertising and marketing by betting and online casino companies reached about $3.9 billion in 2025 [67]. The American Gaming Association reports that the volume of betting advertisements fell for a fourth consecutive year in 2025 and was 27% below the 2021 peak [68], and that sports betting accounted for 0.4% of all television advertising in 2024, below alcohol at 0.5% [69]. The industry presents those figures as evidence of restraint. They also show that the heaviest advertising came in the launch years, when operators were acquiring the customers whose financial outcomes the research in section 3 measures ◈ Strong Evidence.
The audience that matters most is the youngest. A University of Bristol study of gambling advertisements on social media found that 19 of 24 advertisements tested were twice as likely to appeal to people under 25 as to older adults, and that 15 of 24 produced positive emotional responses in children and young people [70]. In the NCAA survey, 63% of students living on campus recalled seeing betting advertisements [48]. Several US universities signed marketing agreements with sportsbooks after 2018. Caesars ended its campus deals with Louisiana State University and Michigan State University in May 2023 after press reporting and protests from faculty [71]. Ontario's regulator banned active and retired athletes from gambling advertisements from February 2024; its chief executive, Tom Mungham, said that “children and youth are heavily influenced by the athletes and celebrities they look up to” [72].
Saturation has also changed the experience of the athletes themselves. About 15 US states, including Ohio, Maryland and New York, now ban bets on the performance of individual college athletes, after reports of abuse from bettors [73]. In a March 2026 NCAA survey, one in three men's college basketball players reported online, verbal or physical abuse related to betting [73]. Integrity cases have followed. In October 2025 federal prosecutors charged the NBA player Terry Rozier in a scheme that allegedly used inside information about players to win bets; the Portland coach Chauncey Billups was charged at the same time in a separate case involving rigged poker games, not sports betting [74]. In November 2025 prosecutors indicted the Cleveland Guardians pitchers Emmanuel Clase and Luis Ortiz on charges of rigging individual pitches for bettors; Ortiz denies the charges [75]. These are allegations, not convictions. But they concern bets on single moments within a game — the products that the parlay strategy described in section 2 has made most common.
Restrictions have arrived country by country, and each covers only part of the problem:
- Britain. Premier League clubs agreed in April 2023 to remove gambling sponsors from the front of matchday shirts from the 2026-27 season; sleeve and perimeter-board sponsorship remains allowed [76].
- Ontario. Athletes have been banned from gambling advertisements since February 2024, except in responsible-gambling messages [72].
- Italy. A total ban on gambling advertising and sponsorship has applied since 2018 and 2019 [77].
- Spain. A 2020 decree limited broadcast gambling advertising to between 1am and 5am; the Supreme Court annulled its online and bonus provisions in April 2024 [78].
- Australia. From January 2027, no gambling advertisements during live sport broadcasts between 6am and 8.30pm, no athletes in advertisements and no logos on uniforms [79].
Six Jurisdictions, One Pattern
Britain, Australia, Brazil, Canada, southern Europe and Japan
The American experiment is recent, but the countries that liberalised earlier show where it leads. Australia now records the highest gambling losses per adult in the world, A$1,528 a year ✓ Established [24]. Britain has spent two decades trying to regulate a market it opened in 2007. Brazil legalised online betting in 2023 and within a year faced evidence that welfare recipients were sending billions to betting companies [80]. Japan has chosen not to legalise at all [26].
Britain's market is the most mature. In the year to March 2025 the total gross gambling yield of the British industry was £16.8 billion, of which online betting supplied £2.6 billion, and football alone £1.3 billion [81]. The government's April 2023 white paper, High Stakes, set out the first comprehensive reform in nearly two decades; the culture secretary of the time, Lucy Frazer, described it as bringing “pre-smartphone regulations into the present day” [82]. Its main measures have since been implemented: a statutory levy on operators, expected to raise about £90 million to £100 million a year for research, prevention and treatment, took effect in April 2025, and online slot stakes were limited to £5 per spin, and £2 for 18- to 24-year-olds [83]. In November 2025 the Treasury went further, raising remote gaming duty from 21% to 40% from April 2026 and creating a 25% rate for remote betting from April 2027, measures forecast to raise about £1 billion a year by 2029-30 [84].
Australia offers the clearest case of a government declining to follow its own inquiry. The Murphy committee made 31 recommendations in June 2023, the central one a comprehensive ban on online gambling advertising [8]. Some consumer-protection measures followed: from June 2024 online wagering providers may not accept credit cards or cryptocurrency [85], and BetStop, a national self-exclusion register launched in August 2023, had more than 67,000 registrations by its third anniversary [86], with about 40,160 exclusions active at the end of June 2026 [87]. But in April 2026, almost three years after the report, the government announced only partial limits on advertising from January 2027 [79], and in May 2026 it formally rejected the total ban. The independent MP Kate Chaney, a member of the inquiry, called the decision “a gift to the gambling lobby” [9].
Losses on legal gambling in Australia were A$33.0 billion in 2024-25, or A$1,528 for every adult; wagering on racing and sport accounted for A$8.45 billion and poker machines for A$17.44 billion [24]. The Lancet Commission described Australia as the country with the highest gambling spending per head in the world [38].
Brazil shows how fast harm can become visible when a large, lower-income population gains access to mobile betting. Law 14.790 of December 2023 legalised online fixed-odds betting, and unauthorised operators became illegal from January 2025 [25]. Before the regulated market even opened, a technical note from the Central Bank found that in August 2024 about 5 million people in households receiving the Bolsa Família welfare benefit had sent about R$3 billion to betting companies through the instant-payment system Pix ⚖ Contested [80]. The industry disputed the figure, arguing that net spending by beneficiaries was about R$210 million [88]. The national retail confederation estimated that retail lost R$103 billion in sales in 2024 as household money flowed to betting [89]. In the first regulated year, 79 authorised companies reported gross gaming revenue of about R$37 billion from 25.2 million bettors, and the authorities blocked more than 25,000 illegal sites [90]. A Senate inquiry into the sector ended in June 2025 when its final report was rejected by four votes to three [91], and Congress later approved a gradual rise in the betting tax from 12% to 15% by 2028 [92].
Canada and southern Europe show different stages of the same cycle. Canada legalised single-event sports betting in August 2021 and left regulation to the provinces [93]. Ontario opened a competitive online market in April 2022; in its third year, 2024-25, wagers reached C$82.7 billion and revenue about C$2.9 billion, with sports betting about a quarter of the total and online casino games the rest [94]. Italy took the opposite approach to advertising, banning all gambling promotion from July 2018 [77]. Serie A clubs say the ban costs them €100 million to €150 million a year in sponsorship, the president of the national football federation called it largely ineffective, and the sports minister has said that work on repealing it has begun [95]. Spain restricted advertising by decree in 2020, but the Supreme Court annulled the provisions on bonuses, celebrities and online advertising in April 2024, ruling that such limits required a law passed by parliament [78].
Japan is the control case. The Penal Code prohibits gambling except on state-run horse, bicycle, boat and motorcycle racing and the toto football lottery, introduced in 2001, whose profits are divided between sport and the national Treasury [96]. In December 2025 a cross-party group of Diet members chaired by the former prime minister Taro Aso agreed unanimously to keep sports betting illegal, and set up a team to combat match-fixing [26]. Prohibition does not mean absence. The same report estimated that Japanese users wagered about ¥6.5 trillion on overseas betting sites in 2024 [26], and a National Police Agency survey found that 3.37 million Japanese had used overseas online casinos, 46.2% had borrowed money to gamble and about 60% of users were in their twenties and thirties [97]. In June 2025 the Diet made it illegal to direct users to offshore gambling sites [98]. The best-known Japanese case of betting harm occurred in the United States: Ippei Mizuhara, the former interpreter for Shohei Ohtani, moved nearly $17 million from the player's account to pay an illegal bookmaker, and was sentenced in February 2025 to 57 months in prison [99].
Across these six jurisdictions the sequence is strikingly similar. Legalisation or liberalisation is justified by the existence of an illegal market and the promise of tax revenue. Advertising expands rapidly in the first years. Evidence of harm accumulates, usually first through helplines, welfare data or academic studies rather than official statistics. A public inquiry then recommends strong restrictions, and governments adopt a weaker version, partly because sport and broadcasting have come to rely on betting money. Japan is the exception that tests the rule: it has kept prohibition, and it still faces a large offshore market and a debt problem among young online gamblers [97]. The comparison does not show that prohibition works. It shows that legalisation, once adopted, is very hard to reshape, and that the moment of greatest regulatory freedom is before the market opens ◈ Strong Evidence.
The Revenue Promise
Taxes promised, taxes delivered and the rise of prediction markets
Sports betting supplies about 0.2% of average US state tax revenue, and only one state, New Hampshire, receives more than 1% of its tax revenue from it ✓ Established [7]. The revenue has grown rapidly from a small base, and states have been raising rates since 2023. But the fiscal case for legalisation was always weaker than its supporters claimed, and a new unregulated competitor is now eroding it.
The first wave of legalisation missed its targets. Of the six states that launched sports betting in 2018, four collected far less tax than projected in the first year: West Virginia received about a quarter of its forecast, and Mississippi and Pennsylvania about half [100]. Supporters of the industry argue that those early figures reflected optimistic launch-year assumptions, and that in mature markets revenue has broadly met or exceeded the projections that states published ⚖ Contested [101]. Both points can be true. Revenue has grown strongly: US Census data show that sports betting tax revenue rose 382%, from $190 million in the third quarter of 2021 to $917 million in the second quarter of 2025 [23]. But growth from a small base does not make the total large. In 2025 states collected about $3.3 billion from sports betting, compared with $8.3 billion from alcohol taxes and $14.8 billion from tobacco [7].
The net figure is smaller still. Some betting spending comes at the expense of state lotteries, which also fund public budgets. An analysis by the Manhattan Institute estimated that Ohio's $209 million in sports betting tax was largely offset by an estimated $160 million fall in lottery revenue [7]. Some states allow operators to deduct the value of promotional bets from their taxable revenue: in 2023 operators saved more than $150 million in tax across six states that permit this, and in Pennsylvania deductions exceeded $228 million [102]. New York, with a 51% tax on mobile betting revenue, the highest in the country, allows no such deduction, and it directed $1.33 billion to education in the 2025-26 financial year, along with $12 million to problem gambling services [103]. That ratio — about one dollar for treatment for every hundred dollars for schools — illustrates how states account for the costs of the product they tax.
Since 2023 the direction of tax policy has been upwards. Ohio doubled its rate to 20% in 2023, and Maryland raised its rate from 15% to 20% in 2025 [104]; and from July 2025 Illinois became the first state to impose a tax on each bet, of $0.25 on an operator's first 20 million wagers and $0.50 thereafter, on top of a graduated tax on revenue [105]. At the federal level, the tax law signed in July 2025 capped the deduction for gambling losses at 90% of winnings, so that a bettor who breaks even can owe tax [106]. Britain's November 2025 budget nearly doubled the duty on online gaming and raised the rate on remote betting [84], and Brazil approved a gradual increase in its betting tax [92]. Governments that once presented betting as a source of painless revenue are now taxing it in the way they tax alcohol and tobacco: as a product whose social costs justify a higher rate ◈ Strong Evidence.
State budgets record the tax receipts from sports betting but not its costs. The New York Fed's finding that neighbouring states bear financial harm while collecting no revenue [17], the lottery losses that offset some of the gain [7], and the $43 million in unspent treatment funds [63] all fall outside the figure that legislators see. A fiscal case that counts only one side of the ledger is not a fiscal case.
A new competitor is undermining even the revenue that exists. Prediction markets — exchanges regulated as financial derivatives markets by the federal Commodity Futures Trading Commission — now offer contracts on the outcome of sporting events that function much like bets, without state licences or state taxes. Kalshi, the largest, recorded about $5.8 billion in trading volume in November 2025, about 91% of it on sports [107], and in September 2026 it recorded a $15.27 billion week and its first $3 billion day [108]. In April 2026 a federal appeals court ruled by two votes to one that New Jersey could not regulate Kalshi's sports contracts because federal law gives the commission exclusive jurisdiction, while courts in Ohio and Massachusetts reached the opposite conclusion [109]. The American Gaming Association says prediction markets have already diverted more than $500 million in potential sports betting tax revenue [4]. Its chief executive, Bill Miller, called the fight against them “a defining fight for our industry” [4].
The rise of prediction markets also removes many of the consumer protections that justified legalisation. State-licensed sportsbooks must verify ages, offer self-exclusion and fund problem-gambling services. Federally regulated exchanges are not designed as gambling operators and are not bound by state gambling rules, so the protections depend on the outcome of litigation that is still under way [109]. The policy choice that states made in 2018 — legalise, license and tax — is therefore being bypassed by a product that offers the same experience under a different legal category. The evidence reviewed below shows that most regulatory approaches tested so far have produced modest effects, and that the most effective measures act on product design rather than on the legal status of betting:
| Policy approach | Evidence of effect | Assessment |
|---|---|---|
| Payment restrictions | Australia's ban on credit cards and cryptocurrency for online wagering removes the most direct route into debt; similar measures target the borrowing documented in US credit data [85] [6]. | |
| Product and stake limits | Britain's slot stake limits and US state bans on college player bets act on the riskiest products directly, though few have been formally evaluated [83] [73]. | |
| Advertising restrictions | Partial limits in Britain, Ontario and Australia reduce visible exposure, but logos and social media continue; Italy's total ban did not stop the growth of illegal betting [22] [110]. | |
| Higher taxes | Higher rates raise revenue and may reduce promotional spending, but they do not by themselves change the concentration of losses among heavy bettors [84] [28]. | |
| Treatment and levies | Treatment funding remains small relative to harm, at an average of $0.54 per person in US states; Britain's levy is new and its effects are not yet measurable [62] [83]. |
What the Evidence Tells Us
The debate, the open questions and the shape of the answer
The evidence does not support the claim that legal sports betting is harmless, and it does not support the claim that legalisation alone caused every harm attributed to it. What it supports is narrower and more useful: the harm depends on product design — mobile access, high-margin bets, advertising and credit — and governments have more control over those features than over the existence of betting itself ◈ Strong Evidence [6] [111].
The first unresolved dispute concerns the illegal market. The industry's argument for legalisation, and against tighter regulation, rests on the claim that restrictions push customers to unlicensed operators. In the United States, the American Gaming Association estimated in 2022 that Americans bet $63.8 billion a year with illegal operators [16], and a 2025 estimate by the firm Yield Sec claimed that illegal operators took 74% of online gambling revenue in 2024, though its methodology has not been published ⚖ Contested [112]. In Britain, a study commissioned by the Betting and Gaming Council estimated that up to £4.3 billion a year was staked with illegal operators [113]. The regulator was sceptical. Andrew Rhodes, chief executive of the Gambling Commission, told MPs that the risk was “overstated” and that he had never been given an operator name or anything else on which he could act [114]. The size of the illegal market is uncertain, and the main estimates come from sources with an interest in the answer.
The second dispute concerns whether problem gambling has actually increased. In Britain, the official survey's estimate of 2.4% to 2.7% problem gambling is much higher than the 0.3% to 0.4% found by older face-to-face surveys, but the Gambling Commission's own research found that much of the gap is explained by the change in survey method, not by a change in behaviour ⚖ Contested [54]. In the United States, the National Council on Problem Gambling's surveys estimated that about 20 million adults showed problematic gambling behaviour in 2024, down from 27.5 million in 2021, a period that included the pandemic [49]. Supporters of the industry cite that decline as evidence that legalisation did not cause a surge [115]. The decline, however, is measured from an unusually high pandemic peak, and the same survey found that risk among young adults and sports bettors remained high [49]. Prevalence surveys and financial records measure different things. The financial records, which do not depend on self-reporting, consistently find harm.
The Case for Legal, Regulated Betting
Industry estimates put illegal US sports betting at $63.8 billion a year before most states legalised [16].
Age checks, self-exclusion and funding for treatment apply only to licensed operators; Brazil blocked more than 25,000 illegal sites in its first regulated year [90].
Sports betting tax revenue rose 382% between 2021 and 2025, and states are raising rates [23] [104].
Most customers bet modest amounts for entertainment, and surveys show problem-gambling estimates in the US fell after 2021 [49].
Japan's ban coexists with an estimated ¥6.5 trillion wagered on overseas sites [26].
The Case That Legalisation Expanded Harm
Legal wagers grew about 34-fold from 2017 to 2025, and household data show the money came from savings [3] [5].
Bankruptcies rose 25% to 30% and credit card delinquency among under-40s rose after online legalisation [6] [41].
Sports betting supplies about 0.2% of state tax revenue, and lottery losses offset part of it [7].
The top 10% of British online accounts supplied 79% of revenue [28].
Searches for gambling-addiction help rose 23% nationally, most where online betting was legal [3].
The third dispute concerns what works. Italy's total advertising ban is the strongest restriction in any large market, yet the illegal online market there is reported at about €20 billion, driven by mirror sites and social media, and legal commentators have described the ban as rigorous on paper and ineffective in practice [110] [95]. No peer-reviewed evaluation of its effect on harm was found for this report ⚖ Contested. The Australian inquiry reached the opposite conclusion from the same kind of evidence: that “partial bans on gambling advertising do not work”, so only a comprehensive ban would [66]. Both positions may be partly right. Advertising bans reduce the visibility of licensed operators, but in a market where betting is available on any phone, they cannot by themselves reduce access. The measures with the clearest logic — banning credit for betting, limiting the stakes on the fastest products, restricting bets on single moments in games — act on the product rather than on its promotion [85] [83] [73].
Taken together, the evidence supports five conclusions about what legalisation actually did:
- It created a new market. Legal wagers grew from $4.9 billion to $166.94 billion in eight years, far more than any estimate of the illegal market it was meant to replace [3] [4].
- It moved money out of savings. Household and credit data show lower investment, more overdrafts and more bankruptcies, concentrated in constrained households [5] [6].
- It concentrated harm in young men. Helpline callers, problem-gambling scores and financial distress are all highest among young men [18] [52] [6].
- It delivered modest revenue. Tax receipts are about 0.2% of state revenue and do not count costs exported to neighbours [7] [17].
- It is hard to reverse. Britain, Australia and Italy show how sport and broadcasting come to depend on betting money, and how inquiries' strongest recommendations are diluted [9] [95].
The 2018 debate treated legalisation as a decision about the law. The evidence since then shows it was also the launch of a consumer product whose most profitable form — mobile, continuous, high-margin and heavily advertised — was chosen by the operators. The harms track those design features more closely than they track legality itself, which is why the most useful policy questions now concern the product, not the permission.
The Lancet Public Health editorial that accompanied the 2024 Commission argued that gambling “is not a simple leisure activity; it is a health-harming addictive behavior”, and that governments should regulate it as they regulate alcohol and tobacco [111]. The evidence assembled in this report is consistent with that view, but it suggests a more specific version of it. Alcohol and tobacco policy works through price, availability, marketing and product rules. Each of these has a direct equivalent for betting: tax rates and promotional deductions; whether betting is available on a phone, at what hours and with borrowed money; whether it is advertised during live sport; and whether operators may sell parlays on single pitches or limit the customers who win. States that legalised after 2018 made these choices quickly, often in the same bill, and the research now shows that they mattered [6] [3].
The final finding concerns who pays. The tax revenue from sports betting goes to state budgets, and in New York to schools [103]. The costs fall on a narrower group: young men in low-income areas whose credit scores fall and whose bankruptcy risk rises [6]; households whose savings and food budgets shrink [42]; partners exposed to violence after a lost bet [15]; and neighbouring states that bear the financial harm without the revenue [17]. Japan's lawmakers looked at this record in 2025 and decided not to legalise [26]. Countries that have already legalised do not have that option. What they have is the evidence of the past seven years, which shows with growing clarity which features of the product cause the harm — and therefore which ones a government could change.