What does football reveal about the impact of Covid-19 on work performance?

Amid concerns about the potentially lower workplace productivity of people who have had Covid-19, evidence from professional football suggests that even six months after an infection, a player’s performance is reduced. Team performance also suffers from more players being infected.

The policy response to the pandemic has been primarily framed in terms of short-term measures – social distancing measures aimed at slowing the immediate spread of the virus; and lockdowns designed to protect healthcare services from being overwhelmed by the particularly resource-intensive nature of treating patients with Covid-19.

The ensuing debate has thus been framed in terms of which policies have been especially effective in reducing case rates, as well as the health costs of the policy response. Both social distancing and, more dramatically, lockdowns may reduce economic output in the short run.

Less commonly discussed, despite increased references to Long Covid, is the longer-term impact on economic productivity following from the health consequences of infection. If those infected cannot function at the same level at work as they previously could, then will they and their workplaces be less productive overall – and will aggregate economic activity be lower?

Are there any useful settings to study this phenomenon?

This is a difficult question to answer as worker performance is hard to measure. What’s more, due to testing limitations, it often not known which workers have been infected with the virus. But there is one particular sector of the economy where insights can be found: professional football.

First, since the initial lockdowns in the spring of 2020, several first-tier European leagues have carried on playing, unaffected by all subsequent lockdowns. Second, players have participated in regular (half-weekly to weekly) testing regimes to ensure that they do not have the virus and spread it. And lastly, football is a very well measured sector of the economy, offering a wide and detailed range of performance proxies. These three characteristics provide an excellent environment in which to understand the longer-term impact of Covid-19 infections on workplace performance.

Official data providers of the professional football leagues (we study Germany’s Bundesliga and Italy’s Serie A) provide information on each event and action in a football match, be it a pass, an interception, a goal or disciplinary action, including the exact location on the pitch and the timing during a match.

Details on the kinds of workplace injuries frequently suffered in football are publicly available (for example, muscle or bone injuries). Taken together with knowledge of the identity of footballers who have been tested positive for Covid-19 since the beginning of the pandemic, it is possible to consider the impact of the virus on productivity both initially and over the longer term.

The regular testing regime means that all infections will be detected, and false positives avoided where a player tests positive but is actually negative. There have been a handful of false positives, which we drop from the sample in our study. Finally, the detailed information on player actions means that it is possible to measure the impact of a Covid-19 infection on commonplace workplace activities, independent of self-reported symptoms.

Does this sector reveal anything about the real world?

Footballers are, of course, a rather distinct set of employees: they are very highly paid and usually retire in their mid-thirties. But as well as this being an ideal setting for observing the workplace impact of Covid-19, there is broader external validity.

Football is a highly demanding physical and mental activity. Many roles in the broader economy satisfy being physically demanding – construction, for example – while others are mentally demanding – such as work in the healthcare sector. The latter is likely to be physically demanding too.

The action of passing in a football match is both a physical and mentally demanding task, requiring players to be at the pace of the broader match taking place over 90 minutes, and able to determine the right passing decision out of a wide range of possible options.

Importantly, also, footballers act within a team, as do most employees in standard work settings. As such, football can reveal lessons about potentially broader impacts on GDP. It can reveal lessons about the impact of Covid-19 infections both on subsequent individual productivity in physically and mentally demanding tasks – and also about the impact on broader team productivity.

How can productivity be measured in this setting?

Our measure of footballer productivity is the number of passes in which they are involved. This is a commonly used measure of labour productivity used in research in labour economics in this specific setting, as it reflects the extent to which a player is involved in the teamwork required for a football team to perform well.

But in our analysis, we first consider the likelihood of playing – whether a player is on the pitch at some time during the match. An important aspect of considering the impact on footballing measures is the selection effect. In football, a player obviously cannot pass the ball if they are not on the field.

What is the impact of Covid-19 infections on productivity for professional footballers?

Figure 1 shows the impact on the likelihood of a footballer playing in a match (the ‘extensive margin’), where the red vertical line indicates when an infection occurred. Points to the left are pre-infection, and points to the right are post-infection. The lighter red area is the 90% confidence interval, the darker red area is the 95% confidence interval.

Initially, players are more than 10% less likely to play after testing positive. This is driven mainly by the quarantine phase. A 10-14 day isolation period is likely to cover at least one match and possibly two or three. Afterwards, players recover to a similar likelihood of playing as before the infection.

Figure 1: The effect of contracting Covid-19 on player selection chances

Source: Authors' calculations

Second, we consider the number of minutes played, since a player may play only a few minutes of a match, despite being named in a squad. Figure 2 presents the impact on minutes played of a Covid-19 infection in a similar manner to Figure 1.

We see a similar time profile, with a sharp drop down immediately after an infection of close to 10%, with some recovery after this (such that the effect disappears over time).

It could be that the same players that were playing less are now playing back at their previous levels, or that Covid-19 players are being selected out of matches. While participation of players already gives some intuition on their performance, it does not provide information on within-match performance – for example, whether players do not play less after an infection because they do not perform worse at all or because they can limit their performance drop to small but still substantial deteriorations.

Figure 2: The effect of contracting Covid-19 on playing time

Source: Authors' calculations

Given this, we turn to what economists call the ‘intensive margin’ – in this case, performances within a match, on the field. We consider the impact on the amount of passing in Figure 3. The impact on passing is less instantaneous, falling about four percentage points immediately, and continuing to fall to more than six percentage points over six months after the initial infection. It seems likely that this more moderated effect is due to the selection effects we have already identified in Figures 1 and 2.

But the significant take-away from this is that even after more than six months after an infection, the number of passes made – a well-established measure of workplace productivity for footballers – is around five percentage points lower than it was before the positive test.

Figure 3: The effect of contracting Covid-19 on passing performance

Source: Authors' calculations

We then consider the in-match variation in performance in passing in Figure 4. For post-infection players, passing performance is always lower than unaffected players, and deteriorates further after 30 minutes on the field. This effect is beyond general performance changes of players within a match.

It is worth noting that passing performance varies not just over time within a match, but that also the Covid-19 effect differs according to different players’ characteristics. For example, older players and players who used to play less before the infection reveal stronger performance drops after an infection.

Figure 4: The effect of contracting Covid-19 on within-match passing performance

Source: Authors' calculations

It is important to look along a number of dimensions, even if passing is a well-accepted measure of the productivity of a footballer. Figure 5 displays a range of other measures.

First, we split passing into short and long passes, and consider the accuracy of passing, as well as ball recoveries, the distance a player covers and how often they intercept the ball during play. The impact on both types of passes is significant, albeit larger for shorter passes. The effect is 3-5%. The effect on pass accuracy is negative but insignificant, as are ball recovery effects. Both the distance covered by post-Covid-19 players, and their number of interceptions, are significantly lower.

Figure 5: The effect of contracting Covid-19 on other performance metrics

Source: Authors' calculations

Are Covid-19 infections different from other injuries or illnesses?

It is important to compare absences due to Covid-19 infections to other types of injury or player absence. It could simply be that players returning from injury perform less well for a significant length of time as they recover and re-adjust back to playing. This may not be unique to Covid-19.

In Figure 6, injury data are used to look at this. Panel A displays general injuries broken into short-term injuries (missing up to two matches) and longer-term ones (over two matches). Panel B shows performance after absences due to respiratory conditions. In both cases, the profile of passing performance is insignificantly different from zero, in contrast to the impact of Covid-19 on passing performance. This might be due to less knowledge of how to treat the disease (mechanical) or stronger health consequences for human airways as found in medical research (physical).

Figure 6: Pass performance effects following different types of player absence

Panel A:

Panel B:

Source: Authors' calculations

Is there a team-level impact?

In the workplace, as on the football field, teamwork matters. Teams are complex organisations, but nonetheless it makes sense to consider whether in the footballing context, it is possible to detect any team-level impacts of Covid-19 infections. We look at the ratio of Covid-19 infected players to total players in a team at any point in time, and look at how that influences the team’s passing performance.

In Figure 7, we split the level of this squad exposure to Covid-19 into four quartiles – from the least affected (first quartile) to the most affected (fourth quartile). We see that while for the first three quartiles, there is no statistically significant impact of Covid-19 on team output, the most seriously affected teams in the fourth quartile are significantly negatively affected, with team passing performance falling by around 7%. Affected teams also win less often, but this effect is not statistically significant.

Figure 7: The impact of Covid-19 infections on team passing performance

Source: Authors' calculations

What does this mean for economic activity more generally?

Taken together, this evidence suggests that Covid-19 may have a long-lasting impact on workplace performance. We study a group of individuals for whom we have extensive information on workplace performance, but also Covid-19 status.

Those who have suffered from Covid-19 infections can be up to 5% less productive, even up to a year after an infection. These individuals’ teams are affected too. Organisations in one of the most competitive sectors of the economy are unable to absorb such adverse shocks,.

With many millions of infections, in some countries reaching around 10% of the population, a sustained fall in the productivity of such a proportion of the workforce may well have a significant impact.

It may also be a neglected factor in the debate surrounding ‘zero Covid’ strategies (the attempt by some countries to eliminate the virus completely, contrasted with looser policy strategies such as that in the UK, where daily cases remain substantially high). If those that have been infected with Covid-19 are less productive, this will have important effects on labour market outcomes and economic growth more broadly.

The sample setting of professional football is that of physically fit individuals in a physically and mentally demanding workplace. Ideally, further study in other sectors of the economy would be carried out to provide a broader evidence base – although as argued above, football is an ideal setting given its testing regime and data richness.

In addition, it is not possible to distinguish between asymptomatic and symptomatic players to analyse their long-run reaction to an infection. Our sample setting is also pre-vaccination, and hence the successful vaccination in many developed nations like Germany and Italy may affect the future impact of Covid-19 infections. If so, then given the imbalanced nature of vaccinations across the world, as well as across different groups within countries (such as by age and ethnicity), inequalities may be exacerbated by the productivity impact of Covid-19.

Where can I find out more?

Aiming for zero COVID-19 to ensure economic growth. This VoxEU column argues that by choosing a swift elimination strategy, several countries around the world have gained control over the virus.
A study on measuring productivity effects during COVID-19: This NBER report highlights the impact of Covid-19 on productivity.
A study on lockdowns and worker productivity: This CEPR report highlights the effects of lockdown on worker productivity.
A study on worker productivity when working from home: This IZA article focuses on cognitive performance in the home office, using evidence from professional chess.