Most arguments about legalized sports betting run on anecdotes. This paper runs on bank data. Scott Baker, Justin Balthrop, Mark Johnson, Jason Kotter and Kevin Pisciotta took transaction-level records from a financial aggregator covering 230,171 US households — roughly 4.9 million household-quarters — and exploited the staggered rollout of online sports betting across the 26 states that launched between August 2018 and September 2023. Because legalization dates were set by idiosyncratic legislative and court schedules, the timing gives them something close to a natural experiment.

What they find is that sports betting does not behave like other entertainment spending.

  • It does not displace other gambling or consumption. The money comes from somewhere else: after legalization, treated households show a relative reduction in net investment deposits of about 14% — including in robo-advisors, the accounts people use for long-term saving.
  • The hit lands hardest on households with the least slack. For financially constrained households, credit card balances rose by about $368 relative to less constrained ones, an 8% increase over the sample mean, with available credit falling and overdraft frequency rising. Spending on complementary entertainment categories rose at the same time — the betting habit drags other spending with it.
  • It is sticky. Conditional on betting once, about 70% of bettors deposit again at least twice more, and nearly 40% do so more than ten times. Bettors were four times more likely to have played the lottery beforehand, so this is less about new gamblers than about existing ones being handed a faster product.
  • Exposure converts. Conditional on living in a treated state after legalization, the probability that a household becomes a bettor is 13.9%.

The framing point the authors make is worth keeping: the industry’s profitability is itself evidence that the typical bettor faces negative expected returns. If the product were beneficial to most customers, the operators would not be reliably profitable. The paper’s conclusion is that access to online sports betting “comes at the expense of equity market attachment and exacerbates financial difficulties faced by constrained households.”

Two honesty notes. First, this is a working paper (NBER 33108, November 2024) that has since been peer-reviewed and published in the Journal of Financial Economics (2026, vol. 183) — cite the published version. Second, the method is difference-in-differences on state rollout: it identifies average effects on households in legalizing states, not the effect on any individual, and the estimates carry the usual caveats about staggered adoption. The direction of the result is what matters, and it agrees with independent work on the same question using different data — a credit-report panel of roughly seven million consumers found higher bankruptcy rates, more debt sent to collections, more debt-consolidation borrowing and more auto-loan delinquencies after legalization, strongest in states allowing online betting and among young men in low-income counties.