By Daniel Walton
With the Tokyo Olympics underway, athletes are gearing up to triumph or falter in the names of their countries. So, the question on a lot of minds is: who will be atop this Olympics? Of course, the obvious answer is the United States, victors of every Games since 1996 and the most powerful economy in the world. The bookies tend to agree, giving the United States an implied probability of 90% to finish on top.
But a corresponding battle is happening in the world of numbers, as maths and economics geeks attempt to predict the final medal tally. You’d think that this might be a complex field, with multiple, complicated micro-level variables being used to describe the most successful nations, however you’d be wrong.
In fact, the most successful models have tended to be the simplest, with Andrew Bernard’s model correctly predicting the order of the Top 5 nations in the 2012 and 2016 Summer Olympic Games. Bernard’s model relies on only four variables: population, per capita income, past results, and a host bonus.
It is not hard to discern why these variables are effective predictors. A large population provides large pools of potential athletes, high per capita income levels mean greater resource capacity to train and develop athletic talent, and host athletes will generally perform at higher levels due to familiarity, crowd effects and greater rest.
Likewise, these types of predictions aren’t limited to the Olympics. Consider the world sport, soccer: The Economist showed that differences in income, population and soccer interest account for 40% of all variation in soccer results over the past 28 years, and six out of seven of the top predicted nations being among the best in the world. The other, the United States, is a relative outlier in its anti-soccer culture and yet is steadily improving in the international standings.
Other sports, like Cricket, are generally dominated by two or three highly interested nations, and thus these predictions don’t have much significance. However, at the aggregate level (e.g., the Olympics), interest differentials should balance out. Thus, I’ve taken it upon myself to attempt to replicate these models and examine how Olympic sports can be predicted from a purely statistical standpoint. Nevertheless, it seems unfair for an economic model to include the past results achieved by a particular country, so I’ve only examined the effects of population, income, and host effects to form my own predictions.
Using what is called a tobit regression, I have produced the following regression model output, showing that there are significant associations between a country’s Olympic performance and population (lpop), per capita income (lgdp) and host country effects (host).
However, applying this model to 2020 data to provide a best guess of this year’s performance provides for some interesting outcomes. Unfortunately for America, for the first time since 1992, we may be looking at a new country atop the medal table, and it’s not who you’d expect:
That’s right, the purely economic data suggests this year’s Olympic victor will be Japan, the host nation! Now, before all you punters slap $50 on Japan, it should probably be noted that the host effect will likely not be as strong as years past, as Japan’s athletes won’t have their countrymen present to cheer them on in the stands. Reducing the size of the home country effect to 25%, we get the following projections:
Further, eliminating any sporting culture effect tends to overrate Japan, who has comparatively underperformed in most sporting events with respect to their economic clout. Japan, despite having a larger population and GDP, consistently performs below Great Britain, Germany, and France in these events. Likewise, countries such as India and Brazil, specialists in sports without much Olympic prestige, are generally nowhere to be seen in the medal count tables.
This effect can be somewhat accounted for by providing a regional input to the model. For example, North American nations play the same kinds of sports as each-other, as do Latin American and European nations. Here binary indicators use southern Africa as the baseline. These effects are, indeed, most pronounced in Oceania, North America, Europe, and South Asia, with significance levels of <0.1%.
This last iteration of the model is somewhat of a return to normalcy, with the US far ahead of its typical competitors in the medal count. However, there are a few howlers, such as Canada rounding out the Top 3. Certainly, don’t expect Canada to perform this well – at least not in the Summer Games.
So which tally should you trust? I’d lean towards the last count; however it seems to unfairly penalise China for being located in East Asia, which typically underperforms in the Olympics. Nevertheless, with bookies paying out $21.00 for Japan (and $5.50 for China), the United States may be getting some undue respect, as certainly evidenced by the embarrassing defeats their Basketball and Women’s soccer teams have suffered. Likewise, with the unpredictability the last two years have shown us, it wouldn’t surprise me if the United States is dethroned this year as the top sporting nation.
Indeed, a poor showing at the games for the US may have political ramifications elsewhere. Olympic triumph is inevitably a hallmark of a great superpower, and China will be keen to demonstrate another area of American decline. Likewise, Republicans would certainly point to a poor showing as a critique of Joe Biden’s presidency, likely blaming it on “woke athletes who kneel for the flag,” which is already the scapegoat in conservative circles.
Of course, UQES doesn’t endorse any betting positions so don’t take all these predictions to heart, but the economic data suggests US Olympic hegemony may in fact be over.
Bernard, Andrew B, & Busse, Meghan R. (2004). Who Wins the Olympic Games: Economic Resources and Medal Totals. The Review of Economics and Statistics, 86(1), 413–417. https://doi.org/10.1162/003465304774201824