You may have seen this recent paper on the financial effects of legal online sports betting floating around econ twitter a few weeks ago:
Gambling Away Stability: Sports Betting's Impact on Vulnerable Households by Baker, Balthrop, Johnson, Kotter, and Pisciotta (2024).
The headline result: legalizing online betting increases betting by about $25 and decreases investments in stocks by about $50 per household per quarter.
These are averages across all households in a state, including the ones that never bet. For households that actually begin betting after legalization, betting expenses rise to about $179 per quarter, on average.
The negative effects on investment are large relative to the sample mean. The average household invests about $380 a quarter so a $53 decrease in investment is a loss of 14%. These negative effects are concentrated among the top few percentiles of bettors and especially the biggest bettors who were already financially constrained by low incomes, debt, and bank overdrafts.
The author’s identification strategy relies on the staggered rollout of sports betting legalization across states after a 2018 supreme court decision which struct down a federal ban. They do a difference-in-difference regression which means they compare similar households in states that legalize to ones in states that don’t both before and after legalization.
This should isolate the causal effect of legalization on betting or investment if state legalizations are random w.r.t the financial health of households in that state. But that might not be true e.g if states are particularly motivated by the extra tax revenue or business that legal sports betting attracts. If the states that legalized are compensating for otherwise faltering financial health, then the author’s regression will attribute this selection effect to legalization.
There’s not an obvious pattern in the states that end up legalizing betting, though 9/10 of the top population growth states over this period did not legalize betting. The authors present some evidence that state financial health doesn’t predict legalization, but I found their presentation of this evidence in table A.5 unclear.
All of the effects the authors cite are comparisons to outcomes in the control states, mapped above in gray. So if the booming labor and housing market of Texas over this period means households in the control states are massively increasing their investments, then we could see a negative effect in the treatment states even if investment in these states is also increasing.
I would like to see the uncontrolled time trends of investment in the control vs treatment states to see if this is the case and if the parallel trends they show in graphs like Figure 7 above hold without their chosen control variables.
In the appendix, the authors also find positive effects of sports betting legalization on quarterly restaurant, entertainment, and travel spending, and positive effects on all other types of spending not including those. They explain these as complementary goods to sports betting, but the effects seem weirdly large to me. They find that sports betting leads to $100 extra dollars per quarter on restaurants (table A.18), $101 extra dollars per quarter on entertainment (table A.20), and $303 extra dollars per quarter of all other spending (table A.21).
It feels like they are proving too much with these magnitudes. It doesn’t pass the sniff test that legalizing sports betting would increase the average household’s consumption by $500+ per quarter on top of the extra money they’re already spending on bets. This suggests other effects causing differences between spending in legal and non-legal states. As far as I can tell, they don’t break this down by whether the person ever bets so it’s not clear whether this extra spending is only coming from people consuming the complementary good of sports betting or not.
Overall, the paper is well-written and thorough despite my missing wish list. This is still a single pre-print empirical economics paper, though. The base rate of replication and robustness for these is not that high, so I wouldn’t update massively on it.
What should be done?
The negative effects of sports betting on other investments are intuitive and credibly supported by this paper. Let’s accept their results and consider a more interesting question: What should be done?
Baker et al conclude their paper saying:
In light of these findings, it is imperative for policymakers to consider the broader financial implications of sports betting legalization. Targeted interventions, stricter regulation of gambling advertisements, and support for safer investment opportunities are crucial to mitigating the adverse effects on financially vulnerable populations.
The claim that their evidence justifies stricter regulation on sports betting is far beyond what they can support.
The authors have strong evidence for people substituting investments in stocks for spending on what they call “negative expected value” investments. This name reflects the fact that sports betting is not a reliable way to make money. But neither is buying movie tickets or going out to eat. Sports betting is not an investment vehicle, it’s a form of consumption.
When a new product comes on the market and people decrease their savings to buy more of this product, that is evidence that their welfare has improved! When Taylor Swift came to Stockholm she decreased the savings of people there relative to those in Copenhagen, but everyone was glad to spend their money and go to the concert.
The authors of this paper say nothing about the consumption value of sports betting, so they have nothing to say about the consumer welfare effects of legalization.
The intuition that the authors implicitly rely on, but do not actually argue for, is that “financially vulnerable populations” overestimate the consumption value of gambling and thus make themselves worse off by consuming it.
This claim directly contradicts the rational agent assumptions of economics, but it definitely contains an important truth.
If I offered you $100 today or $110 in 3 weeks you might the $100 right away; is $10 really worth three weeks of waiting? But if I offered $100 in 10 years or $110 in 10 years and 3 weeks you might be willing to wait the extra bit of time since it’s so far in the future anyways.
Similarly, you might commit to sacrificing your favorite foods and getting up every morning to go running, starting tomorrow. But when the time comes and you’re faced with the immediate choice the costs of getting up and the benefits of eating ice cream are magnified.
This is hyperbolic discounting. A traditional rational agent considers these tradeoffs the exact same way whether they are faced in the next ten years or the next ten minutes. They discount future gains using exponential decay. So a three week wait costs the same whether it starts today or tomorrow. Hyperbolic discounting magnifies the costs and benefits of near term events, just like how holding your thumb close to your face can make it bigger than the moon in your field of view.
Hyperbolic discounting can lead to consistent short-term decisions that conflict with genuinely held long-term preferences. The application to gambling is clear: getting out of the game is necessary for their long-term plans but when betting is easy and immediate, it's hard to resist.
People who want to buck their bad habits in the long-term will often struggle to do so when the time is at hand.
Hyperbolic discounting is a useful mental model for understanding addictive behaviors both personally and academically. I am confident that excessive short-term thrill-seeking drives much sports betting. Whether or not legalizing betting can still be welfare improving under these conditions is a complex empirical and theoretical question that the authors of this paper do not engage with at all.
Baker, Balthrop, Johnson, Kotter, and Pisciotta’s numbers on decreased investments after sports betting legalization is only compelling evidence that betting should be banned if you already thought all betting was a mistake.