Current Record 32-23

We are back (to the excitement and derision of the commenters) for a combined two week update. The quick update is above. We remain cash flow positive by mechanically taking the opposite side of a bet made by a bloviating gasbag. If you want to see the games for this week, they’re at the bottom of the post.

As I have alluded to before, there are a number of academic papers that have come out recently that discuss the problems associated with sports gambling since the 2018 Supreme Court decision. I want to walk y’all through one of them today. The paper is “The Financial Consequences of Legalized Sports Gambling” by Hollenbeck, Larsen, and Proserpio. They study the effect of consumers’ financial health as measured by average credit scores in states that have legalized sports gambling. Since 2018, they find a substantial increase in bankruptcy rates, debt collections, debt consolidation loans, and auto loan delinquency, as well as restrictions by financial institutions on access to credit. Furthermore, they find that this effect is being almost entirely driven by online access. Not great!

So how do they come to this conclusion? First, they document that there is a lot of gambling. Like a lot a lot. Between 2018 and 2023 nearly $300 billion has been wagered via new legalized sports gambling markets, and most of that comes from online betting. Most of the time, gambling is a form of recreation, but we know that some percentage of gamblers are “problem gamblers” who have difficulty with gambling affecting many aspects of their health and well being. These are the likely folks who are most affected by changes in sports gambling laws because we also have a good idea that problem gambling varies across demographic groups. Younger folks and folks who less disposable impact tend to be more vulnerable to the effects of gambling on financial health, but also psychological profiles like individuals with higher impulsivity might be risk factors. The fancy research words for this is “heterogeneous populations” which if you ever read just means “not everyone is the same.”

The researchers argue that legalized sports gambling caused these effects on average credit. They do this by using a research design called staggered differences-in-differences. In short, they look at states who have legalized sports gambling since 2018 and compare them to states that have not during that period. An additional quirk is that not all the states that legalized sports gambling did so at the same time, what is known as “staggered adoption.” For us to believe this, we have to believe that individuals in the states that did legalize online sports gambling (fancy research word is “clusters”) would have had the same financial health outcomes as the individuals in the states that did not subject to appropriate controlling for unobserved characteristics. We call this the “parallel trends” assumption, and if we do not find it plausible then we do not buy the study. For what it’s worth, the authors do some work to probe the assumption, though I think they probably should do more (it is a working paper after all) before I would fully buy this design.

The primary data source is the University of California Consumer Credit Panel, a dataset that contains anonymized individually records of a nationally representative 2% sample of US adults with a credit report from 2004 to the present quarterly, and contains demographic information for each data point. The dataset (importantly) includes information on opening and closing of accounts for each individual at the quarter level. In other words, the researchers can tell whenever someone in the data is delinquent on an account. Here’s the summary table of their outcomes.

A couple of things to note here. These effects are mostly small but reasonably precise if we buy the model, and they are the average treatment effect (online gambling!) on the treated (on folks in those states with online gambling!).

The researchers break out the effect using graphs. The way to read these is that the dots represent the point estimate (the probability of bankruptcy goes up by 0.01%) and the bars represent the confidence interval. If you see these graphs, the short way to read them is that if the bar doesn’t cross 0 the effect is “statistically significant” on the outcome.

Effects are mostly driven by men more than women and low income men (the 5th bars in these graphs) more than high income men in the sample with the caveat that the bars (how imprecise our guesses of the effect are) are quite wide for this group relative to others.

So what’s the takeaway for policy? For me the obvious one is that gambling like any vice can have negative financial effects for folks, that if we buy the model they have increased since the advent of online gambling, particularly for low income men. The ease of online gambling relative to traditional gambling (for which the overall effects are a fairly precise null) suggests that the reduced friction in placing bets is a likely mechanism for these negative effects. Finally, if you or someone you know has psychological factors that increase gambling risk, it is worth paying attention to their gambling habits.

Picks

This week taking the opposite of Clay’s bets means that we have:

• Virginia Tech +19.5
• Oklahoma State +4.5
• Minnesota and Michigan Over 36.5 (likely all of which to be scored by Michigan
• Ole Miss -18
• Oklahoma -2.5 and the Over 45.5
• Arkansas +3.5 and the Over 52.5
• Penn State -19 and the Over 47.5
• Georgia -2.5 and the Over 48.5
• South Alabama and LSU Under 65.5
• Oregon -25.5