Over 100,000 people have died in the UK in under a year from Covid-19, so it may seem a little crass to talk about sport and Covid. But sport matters economically, it matters physically, and it matters mentally.
A significant part of the economy revolves around our leisure choices – what we do when we’re not at work. Those choices we mean that a lot of people are employed in the leisure sector. Gyms, leisure centres, national parks, 5-a-side football centres, go karting tracks. Sports clubs, football clubs, social clubs, pubs, and the broadcasting industry. Sport and leisure are big business. It matters economically.
Sport matters physically, and mentally, too.
The UK government has estimated that the broader cost to society of obesity and weight-related illness is £27bn a year, and there’s a well-established link between physical and mental health.
Sport matters economically, physically and mentally.
Sport and Covid – what do we know? I’m going to argue that we know that: Sporting events can be super spreaders. Sports fans respond to Covid-19. Sporting outcomes in a time of Covid are different.
First, can sporting events be super spreaders?
Sporting events are economic events, but they are also social events. Since March 2020 we’ve been reminded of the risks that that social events pose in a health crisis. Most sport is not amenable to social distancing – even if my own football team, Oldham Athletic, defends like it is social distancing.
We compete at close quarters, and we watch huddled together.
Until last March, we never really thought about how social events might burden public resources like healthcare systems.
Or did we? In 2016, three US economists, Charles Stoeker, Nicholas Sanders and Alan Barreca, wrote in the American Journal of Health Economics that “Success IS something to be sneezed at”, as they found that US counties that had a team participating in the Super Bowl had an 18% increase in influenza deaths amongst the over 65s.
Fellow American economists Alex Cardazzi, Brad Humphreys, Jane Rusedski, Brian Soebbing and Nick Watanabe found that US cities with a sport franchise have had between 4 and 24% greater influenza mortality compared to cities without one over the last two decades.
Considering Covid-19, in mid-2020, Alexander Ahammer, Martin Halla and Mario Lackner of the Johannes Kepler Universitaet in Linz looked at NBA basketball and NHL ice hockey events in the US in early March before the league suspended.
They found that each additional event increased the number of Covid deaths by 9% in the areas around venues.
NBA and NHL are indoor mass events, and this evidence from the US has helped form a picture that we are now familiar with: indoor events are particularly effective at enabling the spread of Covid-19.
With a PhD student of mine, Matthew Yeo, and Matthew Olczak of Aston University, I looked at the impact of football matches in England in March on the spread of Covid-19 in April.
Unlike NHL and NBA games, football matches are outdoor events.
Most evidence appears to be associated with indoor events, but nonetheless, football matches like the Liverpool vs Atletico Madrid match on March 11 stand out in the public consciousness.
A lot of football takes place around England, usually. Of the 313 local areas in England and Wales, 247 have football taking place on a regular basis in the top eight tiers of English football.
That is, the Premier League of Manchester United, Arsenal and Liverpool, all the way down to the Northern Premier League North West Division of Marine, Prescott Cables and Pontefract Collieries, and the Southern League South Division of Basingstoke Town, Thatcham Town and Cinderford Town.
In February 2020, 909 matches took place in these competitions, and in the first half of March, 340 matches took place across 188 local areas around the country, and 1.6m people attended these events.
Such a number of mass outdoor events spread across the country allows us to consider the extent to which these events can be associated with an increased prevalence of Covid-19 in the areas that clubs involved in the matches are located in.
It’s important to emphasise that this is an exercise in correlations. Co-relations. Things that happen together. A number of football matches, and a number of Covid-19 cases and deaths in the local area. But are they linked? There could be alternative explanations. Areas with football venues may be sufficiently different to areas without them such that we would expect more cases and deaths. Such areas are likely more densely populated, more ethnically diverse, and have lower income levels. These are all factors known to be related to the spread of Covid-19.
So it’s important to control for these characteristics of the local areas that football clubs are in. By doing this, we try to make all areas more similar, such that the only thing that differs is the number of matches. All students of economics are familiar with the words ceteris paribus: holding all else constant. Once we’ve held everything else constant, we can more reasonably make the case that our relationship could be causal.
We found that the existence of matches in March was consistent with 6 more Covid-19 cases, and 2 more Covid-19 deaths, per 100,000 people. We found the effect was remarkably consistent whether a stadium was full, half full, or nearly empty.
That is, mass outdoor events appear to have an impact on the spread of Covid-19 in the Spring in England.
This finding was contrary to what we expected – or hoped, for that matter. It contrasts with the findings of US-based economists Dhaval M. Dave, Andrew I. Friedson, Kyutaro Matsuzawa, Joseph J. Sabia & Samuel Safford, who look at the Black Lives Matter protests in the US last summer.
They found no impact of the protests, but were able to provide an explanation via mobile phone data – by protesting in the open air with face masks on, the protesters were engaging in lower Covid risk activities than, for example, eating in restaurants or drinking in bars.
Our unexpected finding provoked us to try and find an explanation. While all football is open air in England and Wales, not all aspects of the football match experience happen outside.
Fans pack on buses, and into cars, to get to games. They enter the stadium through turnstiles and indoor areas of stadiums, areas that are often packed with fans mingling. They buy food and drink, visit the bathrooms either in a club house at a non-league match, or under the stands at league matches. When leaving the stadium, fans often crowd through narrow enclosed spaces to leave.
There’s a lot of congregating in enclosed spaces even though football matches are outdoor events and it may be these areas that facilitate the spread of the virus.
I don’t think this means we can’t go back to football matches, or sporting events more generally. But it does appear to suggest that, at least for the time being sporting events will be different, as caution needs to be exercised – not unlike how going to the office is very different at the moment.
Sport and Covid – what do we know? Sporting events can be super spreaders.
Need they necessarily be? Can’t events be better organised to avoid pressure points? Haven’t we adapted our own behaviour since March? Have sports fans responded to Covid?
As attendance figures are well recorded for sports events, we can look at the impact that the pandemic has had on attendance numbers. While most events currently are taking place without fans, there was the anticipation effect last Spring, as news filtered in from China at first, then closer to home in Italy.
With Carl Singleton, a fellow economist here in Reading, in the European Sport Management Quarterly journal, I looked at stadium demand in those first few months of 2020 across the elite leagues of Europe. We found that in Italy, England and Germany, stadium attendances were negatively affected by the previous day’s case and death numbers, but there wasn’t a similar effect in Spain or France.
A limitation on such evidence is the way in which attendances are announced at football matches – it’s tickets sold, rather than actual attendance, that tends to be announced and reported.
Hence announced attendances figures do not necessarily give an accurate sense of attendance patterns. Dominik Schreyer and Nicolas Frevel, of the Otto Beisheim School of Management in Germany, found that after the 2015 Paris terrorist attack on a football match between France and Germany, the number of people who bought tickets for German domestic football matches but then did not attend, increased by over a thousand. This effect lasted for two weeks after the terrorist attacks.
Sports fans respond to salient threats to public health, but we might not be able to tell that from headline attendance numbers.
It may be that smaller leagues provide insight. With Carl Singleton and Dominik Schreyer, in Applied Economics Letters, we looked at matchday attendances in Belarus. As Covid-19 spread around the world into March, and as every professional league in Europe was suspended, Belarus carried on playing and even encouraging fans to attend as normal.
We found though that attendances at the stadiums were much, much lower in the early stages of the pandemic, suggesting that people were reluctant to attend mass attendance events given the threat of the virus. Average attendances on the first three matchdays of the 2020 season were a third of that the season before. This effect was short lived, however, as by late May attendances had recovered to much of their previous level.
The evidence then from Germany and Belarus, is that the response to high profile and salient events may be significant – but short lived.
Closer to home, grassroots football clubs below the National League North and South in England were allowed to admit fans in September. How did attendances look at this level in September through to December, relative to the pre-pandemic period?
Controlling for variation at league level, at different clubs, and proximity to league football clubs, on average attendances were almost 10% higher in the autumn of 2020 than during the 2019-2020 season.
This is consistent with the Belorussian and German results in that after some time had passed, fans felt less need to stay away and socially distance. It could also reflect pent-up demand for football in the flesh, given that none had happened since early March.
Sport and Covid – what do we know? Sporting events can be super spreaders, yet sports fans have responded to Covid-19. But what about the events themselves, and their outcomes?
Sporting events are taking place almost all of the time, even during Covid, generating outcomes.
My Flashscores app tells me that on February 10 alone there was 319 football matches, 294 tennis games, 153 basketball matches, 78 hockey games, 1 cricket match, 52 horse racing events, 5 snooker games and 633 table tennis games. All around the world. Huge numbers of events, each one generating data, observations.
Regional economic statistics, such as those produced by the Office for National Statistics, are hugely useful, and yet at the same time, are limited. They cover sizeable regions, are usually available on an annual basis, and don’t go back that many years. But there’s numerous football clubs, rugby clubs, cricket clubs in most local areas around the whole country, and they play matches on a regular basis over much of the calendar year.
These events have been taking place on a near weekly basis for well over a hundred years, too.
The personnel involved, the ownership structures, the results.
On the field, and off it.
Rob Simmons, Babatunde Buraimo and Stefan Szymanski have charted the emergence of the north-south divide, for example, using football club accounts.
Such data enables us to understand more about economic and social behaviour. There have been some great examples of this over the years.
The Spanish economist Ignacio Palacios-Huerta used penalty kicks to show a famous game theory result – namely that people do play mixed strategies and do so such that success probabilities are equalised across different possibilities – shooting or diving left or right.
Palacios-Huerta also looked at the awarding of injury time at the end of football matches, using the information widely available on the number of minutes added at the end of the game.
Injury time is determined at the discretion of the referee, and should reflect the number of stoppages in play in the match itself.
He found, with fellow economist Luis Garicano, that referees added more stoppage time in matches where the home team was losing by a goal, and the crowd was larger.
They attributed this to referees being influenced by the crowd – social pressure.
The influence of the crowd, then, has had a significant impact on match outcomes.
As well the injury time effect, Liverpool John Moores University researchers Alan Nevill, Nigel Balmer, Mark Williams in the Lancet found that referees were more likely to call fouls against the away team if they could hear the crowd noise.
Covid-19 has afforded scholars an ideal opportunity to investigate the influence of the crowd, since the crowd has, with very few exceptions, been entirely absent from football matches since the start of the first lockdowns around the world in March.
Investigations into the impact of playing without fans are interesting for many reasons. Sports fans care about outcomes, especially when it involves their own team.
But there’s broader interest.
Could it impact overall outcomes from a season? Promotion, relegation, prize monies. It matters also for our understanding of behaviour, too. It’s not unreasonable to think that sports players act differently if they are playing in a full or empty stadium.
But just how differently? And which players? A number of studies have been carried out looking at football without fans.
Carl Singleton, Dominik Shreyer and I wrote one looking at the matches prior to the Covid lockdowns that had taken place without fans.
Prior to Covid, football without fans was a punishment for the home team – usually, but not exclusively, for poor fan behaviour.
The tiny number of such matches has always inhibited any analysis of the impact. Since 2002 in elite European football there had been fewer than two hundred matches played without fans, out of well over twenty thousand matches.
We noted back then that while the effect on match outcomes wasn’t clear (it wasn’t significant and was fairly small), the impact on yellow cards was strong, and robust.
Referees gave almost half a yellow card less to visiting team players in matches without fans.
Since football returned in the summer, after the Covid lockdown, Carl and Dominik and I were joined by Prof Peter Dolton of the University of Sussex, and Prof Alex Bryson from UCL, and we published in Economics Letters an update of our study for the post-lockdown summer matches around Europe.
Rather than a couple of hundred matches without fans, we now have thousands of them.
Yet we found almost exactly the same pattern: a small effect on outcomes (about 3%), with the home advantage remaining, and a significant effect on yellow cards.
We have been joined by a number of others investigating the phenomenon of home advantage when fans are not present.
The picture is that the variation in home advantage without fans is quite substantial across leagues, yet taken together it’s still only about 3%.
Our attention is naturally drawn to the seemingly obvious cases where home advantage has disappeared.
In the Premier League more matches (90) have ended in away wins than home wins (83) after almost two thirds of the season.
Sometimes relatively small samples can mislead.
As the effect is only really strong on yellow cards both before and after the lockdown suspension, we are left to believe that the impact, as other scholars have found, is through the referee.
It turns out that at the University of Reading, we have a top referee in our Psychology department – Daniel Lamport.
He’s actually refereeing Tottenham vs Bristol City in the Women’s Super League right now.
With Dan and another academic referee, Dane McCarrick, a PhD student in Leeds when he isn’t officiating in the National League North and South, I’m now looking further into this mechanism – not least because since 2006, referees have been mic’ed up – not too dissimilar to my own mic currently.
That arguably masks any sound coming from the crowd.
Sport and Covid – what do we know? Sporting events can be super spreaders. Sports fans responded to Covid-19. Sporting outcomes in a time of Covid are different.
These patterns have shaped what sport has looked like since March 2020, and will continue to have an impact. Events are still being disrupted, with great uncertainty remaining about the Tokyo Olympics.
What does the future hold? Economists are known for their awful forecasts, and economic forecasts are especially bad in times of great uncertainty.
Given that, I’ll try and focus on what things won’t change. We still love sport, and desperately yearn its return. Surely we’ll still want to play, and watch. But can sporting structures, that rely on income sources that have been reduced or completely cut off by Covid, survive? Can grassroots structures that rely on volunteers often drawn from vulnerable groups survive?
It’s easy to be despondent, and to worry. But that long historical run of sporting data goes back over a hundred years.
It goes back past the last global pandemic of a scale similar to Covid-19, that of Spanish flu in 1918 and 1919.
In 1921 and 1922, average attendances at top football matches in England hit 30,000, a level that had never previously been achieved.
Sport, and doing social things, was thriving not that long after the pandemic. The roaring twenties.
The research presented here doesn’t preclude fans returning to sporting events soon, if well organised. Will the experience be lessened? I don’t think necessarily – measures to reduce queuing and waiting around are surely great for everyone. That outcomes are affected by the absence of fans is merely another edge – another edge to be used and exploited in the fierce competitive heat of sport.
Sport and Covid – what do we know? Sporting events can be super spreaders. Sports fans responded to Covid-19. Sporting outcomes in a time of Covid are different.