States Bet on Gambling Taxes. The Money Barely Shows Up

Half of American men under 50 now carry a sportsbook in their pocket. Last year they and everyone else pushed $167 billion through the legal betting market, and the apps handed roughly $17 billion of it back to the industry. Somewhere in that torrent sits the money that convinced 38 states to legalize sports gambling in six short years. The promise was simple: let people bet, and the tax revenue will roll in.

A new Manhattan Institute report picks that promise apart, number by number, and argues the revenue was never the windfall anyone was sold. Its author, policy analyst Charles Fain Lehman, is not neutral about gambling, and the report says so plainly, but the arithmetic is worth processing.

Start with the size of the pot, because everything else follows from it. Across all fifty states, sports-gambling taxes brought in about $3.3 billion in 2025. That sounds enormous until you set it beside the $1.5 trillion those same governments collected in total. Gambling raised less than alcohol taxes, less than tobacco, and about the same as marijuana, which is legal in sixteen fewer states. Run the ratio and sports betting accounts for roughly 0.2% of state tax revenue, on average. Only one state, high-tax New Hampshire, clears even 1%. The lopsidedness is starker still per dollar wagered: states capture around a quarter of every dollar spent on lottery tickets but under a cent of every dollar bet on sports, so shifting play from one to the other is a poor trade for the treasury even before any harm is counted.

So the headline number is small. The report’s more unsettling claim is that a good chunk of it may not be new money at all.

The word economists use is cannibalization, and the intuition is homely enough. People have a fixed amount to lose. A dollar bet on a sports app is a dollar not spent on a scratch ticket or a video-poker machine, and if the state taxes those other habits too, the shiny new revenue stream can be draining an old one. The most systematic look at this so far, a 2025 working paper by Poet Larsen and colleagues, tracked lottery sales across seventeen states and the District of Columbia and estimated that legal online betting cut them by 5.4% to 6.2% in the first sixteen months. The cumulative hit topped $100 million. The concern is not new to this study, either: economists have documented gambling products eating into one another for years, from casinos diverting play from existing casinos to national lotteries hollowing out older betting pools, which is part of why the finding is treated as a recognized pattern rather than a one-off.

Here is where a careful reader has to slow down, though, because that $100 million is not spread evenly. Almost all of it comes from New Jersey and Pennsylvania; most other states more or less break even, and the average effect across the dataset is, by the authors’ own reckoning, hard to tell apart from zero. Cannibalization is possible, in other words, not automatic. One study of Pennsylvania alone found no short-term sign of it. Another, from researchers at New York University, looking at two thousand convenience stores, found that online betting nudged lottery sales up by around 17% a month while retail, in-person betting pulled them down by around 16%, and where both launched together the two roughly cancelled. The picture is mixed, and the report, to its credit, does not pretend otherwise even while it leans on the gloomier end.

The Costs That Land on the State’s Desk

If the revenue is small and possibly borrowed, the report’s third move is to ask what legalization costs. Here the literature is thicker, and grimmer. A run of recent working papers, most of them quasi-experimental and most not yet peer reviewed, links the arrival of online betting to a jump in bankruptcies of around 25%, a rise in credit-card delinquencies, a 79% surge in calls to a problem-gambling helpline, and a measurable uptick in screened-in child-maltreatment investigations. The bankruptcy and delinquency figures trace to a widely cited analysis by researchers at UCLA, USC, and Harvard, which pegged the jump in bankruptcy filings at roughly 25% when a state moves from in-person to mobile betting, alongside comparable rises in credit-card and auto-loan delinquencies. One analysis estimated an increase in suicide among young men aged 15 to 34, partly offset by a decline among older men. These are associations drawn from natural experiments, not verdicts from a courtroom, and the report is careful to flag how many of them ride on unpublished papers. But they point one direction.

The clever part of the argument is that these harms do not have to be large to matter. They only have to be larger than the revenue, and the revenue, remember, is tiny. If a single harm costs a state $10,000 a head and touches ten thousand people, that is $100 million: a rounding error against a state budget, but a serious dent against a few hundred million in gambling receipts. The report walks through Ohio to show the sums. The state pulled in $209 million in sports-betting tax last year. Against that it stacks an estimated $134 million to $182 million in added medical and child-welfare costs, plus up to $39 million in lottery losses, and lands on a total that could swallow the revenue whole.

How Much Weight the Ohio Number Can Bear

That Ohio figure is the report’s rhetorical hammer, and it is also where the caution has to be loudest. Lehman builds it by stacking assumption on assumption: a guessed rise in problem gambling from 2.8% to 3.3% of Ohioans, multiplied by a share of gamblers on Medicaid, multiplied by a range of comorbidity costs, and so on down the chain. He says so himself, calling the whole exercise back-of-the-envelope and warning that the figures are orders of magnitude, not precise sums. Nudge any one assumption and Ohio can flip from net loser to break-even. It is an illustration of how the math could go, not a measurement of how it did.

The helpline numbers are a good case study in why the caution is warranted. Calls do spike sharply after legalization, but state health agencies have found that much of the increase is people phoning the number printed in betting ads for technical help, or newly aware bettors finally seeking support that always existed, rather than a clean count of new addiction. The National Council on Problem Gambling itself cautions that helpline traffic should not be read as a proxy for how many people have a gambling disorder. The signal is real; it is just noisier than a raw percentage suggests.

And there is a wider tension. The report’s executive summary speaks with real confidence, declaring flatly that social costs comfortably exceed revenue. Its body is far more hedged, thick with the language of possibility, of reasonable assumptions, of what may rather than what does. That gap between the cover’s certainty and the chapter’s care is common in advocacy writing, and it’s exactly the sort of thing worth watching for when a think tank with a stated position on gambling is doing the counting. The underlying economics is real; the framing leans harder than the numbers strictly allow.

Where it leaves policymakers is the interesting bit. If sports-betting revenue is small, partly illusory, and plausibly offset by costs the state ends up paying anyway, then the usual objection to tighter regulation, that it would starve the budget, mostly evaporates. States could raise gambling taxes, or ban the riskiest bets, and lose very little they were actually banking on. The report’s own preferred conclusion is that gambling taxes should be set to hold down harm rather than to raise money, since they are bad at the second job regardless.

Whether legislatures will read it that way is another question. The revenue argument has a way of surviving contact with the evidence, precisely because a line item, once written into a budget, is hard to strike. But the next time a governor promises that legal betting will pay for the schools or the reservoirs, it’s worth asking how much of that money was ever really going to turn up.

Frequently Asked Questions