TLDR: China, one of the world’s top air polluters, is home to Chinese Super League (CSL) soccer, where each team plays its matches outdoors, exposed to the elements - including all the elements that make up common air pollutants.
Just how might these air pollutants impact the attendance at these CSL matches? And what might this say about the broader impact to all human activity in light of increased air pollution?
The answers are below.
While the blog’s name is Sports and Economics (in that order), let us not forget that the results driven from sports data may just have implications for the wider world around us. As I have argued in the past, a concept such as defensive decision making is a human phenomenon that occurs in real life but can be verified using the Major League Baseball's (MLB's) infield-fly rule, or that we do not need to design and implement complex studies to observe risk aversion but instead simply watch golfers when they change their strategy based on putting for birdie versus bogie. And while we can use the easily accessible sports data to test consumers behaviour in light of bad publicity by way of MLB attendance and steroid revelations, we can also use sports data to evaluate even more general real-world issues, such as the impacts of air pollution on human activity.
But first, let me kick things off with a bit of background before we do such a thing.
China, one of the world’s top air polluters, is home to a brand of soccer, the Chinese Super League (CSL), which is ranked about the same as the top North American league, Major League Soccer (MLS). Both leagues have similar global rankings and experience a similar number of ticket sales per game. However, unlike a handful of MLS-employed stadiums, all CSL teams play in open roof stadiums - exposed to all the elements.
Elements including nitrogen and oxygen - which happen to form NO2 (more on this in a minute).
I want to build a model to predict the CSL attendance to evaluate the impact of the air quality experienced in China. This is motivated by an interesting study that had taken a different approach while using older data and ultimately found no effect. In order to build my own model, I first downloaded the 2014 to 2019 CSL attendance from FBref.com. Then, through the help of my colleague and interpreter, Chloe Sun, I was able to download air quality data from the Chinese Ministry of Ecology and Environment (MEE). Next I build the model to predict the attendance, controlling for information about each soccer match including the identities of the home and away teams and the prominence/skill thereof, day of the week, month, year, yada, yada, yada.
Now, despite this being a very different sport in a very different market, I employ a similar model for the CSL data as I have done in the past for MLB, with equal success: while the MLB models typically explain about 80% of the variation in the data, the CSL model explains 73% of the variation.
But the most important part of this model is the additional controls for the air pollution. Each day, the MEE provides a list of major pollutants when they begin to appear in levels beyond thresholds that could impact human health. These pollutants include Nitrogen Dioxide (NO2) or Particulate Matter less than 2.5μm or less than 10μm (PM2.5 and PM10 respectively). While these warnings from the MEE do not imply imminent death, those particularly sensitive to decreased air quality, i.e. asthmatics, can experience exacerbated respiratory conditions and all individuals may even experience some inflammation of their airways.
When I include indicators for matches that occurred on days with warnings for NO2, PM2.5, or PM10, I find some startling results that I am sure the league organisers wish I had plucked out of thin air:
The results indicate that on days where NO2 pollution is particularly pronounced, the soccer match loses about 7.2% of the attendance it would otherwise see. The yellow lines show the 95% confidence interval band indicating this effect is statistically significant (i.e., the yellow lines do not intersect the x-axis).
Similarly, all else equal, the soccer match loses about 4.9% and 4.5% of attendance for days with PM2.5 and PM10 pollution, respectively (however, note that the yellow 95% confidence interval spans zero for PM10 which indicates that there is not enough evidence to say that this effect is any different from zero).
Taking a step away from the specifics of soccer, this estimated effect due to pollution may have a more general implication: what if this 7.2% loss in ticket sales also implied that days with high levels of NO2 pollution require the same share of the entire population to stay indoors for health concerns? This would mean that one in fourteen individuals (in China) are impacted by NO2 - and that is even before considering that paid attendance often underestimates the true impact of negative demand shocks!
But for all the impact this air pollution has on attendance or the general population, an obvious solution would be to focus on limiting the number of days with adverse NO2 pollution. So let's ask, over the course of the sample, how is each Chinese city with a CSL team trending with number of days of NO2 pollution?
From 2015 to 2019, the average city added 8.4 days with NO2 warnings. And poor Wuhan - it led the way with the greatest number of NO2 days all before becoming the most googled Chinese city and not for tourism reasons.
While sports provide entertainment to a select portion of the overall population, the data and patterns therein provide unique insight into the human condition. And if this study is to tell you anything, it's a bit of a tough time for those humans with respiratory conditions.
If nothing else, these types of analyses can provide valuable insight to policy makers as to how large of a population their decisions could affect for the better.