How sport can make economics better
Or: An excuse to talk about my favourite football club
Today is the last day to complete the reader survey! The survey has only fourteen questions and should take less than ten minutes. As a small thank you, five respondents will be randomly selected to receive a free one-year paid subscription, worth $75. You can fill in the survey here.
Almost a month ago, former Arsenal left-back Kieran Tierney curled the ball inside the far post in stoppage time of Scotland’s World Cup qualifier against Denmark. An entire country erupted in joyous delirium. For Scotland, it meant a long-awaited return to the World Cup. For me, it brought back a memory that is still painful in Dublin and still profitable, oddly enough, in South Africa.
In November 2009, in a playoff between France and the Republic of Ireland for a place at the 2010 World Cup, a high free kick floated into the Irish box in extra time. Thierry Henry controlled it, only not with his foot. He handled the ball, in fact twice, before squaring it for William Gallas to score. The referee missed it, the goal stood, France went to South Africa, and Ireland stayed at home. For Irish supporters this was sporting injustice of the purest kind. For South Africa, Henry’s left hand turned out to be an invisible subsidy.
A few years ago, together with Maria Santana-Gallego, I tried to measure exactly how large that subsidy was.1 We wanted to ask a question that is almost never asked about mega-events: not whether the World Cup boosts tourism in general, but how much it matters which countries qualify. Thirty-two teams travel to the World Cup (next year it will be 48), but they do not bring equal numbers of fans, nor equal media attention, nor equal purchasing power. A France at the World Cup is not the same thing as an Ireland at the World Cup, at least from the perspective of hotels, restaurants and tour operators in Cape Town and Durban.
To answer that question we built a long-run dataset of tourist arrivals to South Africa from every country in the world from 1995 to 2013. We then estimated a gravity model of tourism, the workhorse framework that trade economists use to explain why some countries trade more with each other than others. Richer and larger countries, closer in distance and with historical ties, tend to send more tourists; political stability matters; so do various other frictions. Once these usual suspects are controlled for, we can ask what happens to tourist arrivals when a World Cup takes place, and in particular what happens when a country qualifies for the tournament.
We arrived at two main results. First, the 2010 World Cup did indeed raise tourist arrivals to South Africa, especially from countries whose teams qualified. The effect is surprisingly large. For many participating countries, arrivals in 2010 were 20, 50, even 100 percent higher than in the years before, and there is still a positive legacy effect in the three years after the tournament. Crucially, most of this gain came not from South Africa’s traditional tourism markets – the United Kingdom, Germany, the Netherlands – but from new ones. Honduras, Chile, Mexico, various West African countries and parts of Eastern Europe suddenly discovered South Africa, and many did not forget it again.
Second, the model allows us to run counterfactuals. We can ask, for example, what would have happened to South African tourism if Egypt, not Algeria, had qualified from their playoff; or Russia instead of Slovenia; or Ireland instead of France. Because we know how many tourists Egypt and Algeria normally send, and how tourists from participating countries behave in World Cup and post-World Cup years, the counterfactual is not guesswork; it is an extrapolation from observed behaviour.
When we ran those simulations, the results were sobering. In the Egypt–Algeria playoff, the smaller and poorer country qualified, and South Africa lost out on thousands of higher-spending Egyptian tourists. The same was true when Slovenia beat Russia. In both cases, one late goal in a cold European stadium showed up, years later, in the tourism statistics of a country at the other end of the world.
The Henry incident was different. Our estimates suggest that France’s qualification, courtesy of that handball, led to roughly 28 000 more tourists over the event year and the three years after than Ireland would have done. In 2015 prices, that is roughly R878 million in additional tourist expenditure and more than 6 000 extra jobs in South Africa’s tourism sector. One missed handball in Paris changed employment prospects in Polokwane.
Once you see football in that way, as a series of small shocks with measurable economic consequences, it is hard to stop. That is where Ignacio Palacios-Huerta’s marvellous new survey, The Beautiful Dataset, enters the story.2 He argues that sport is not just a good source of metaphors for economics; it is one of the best environments we have for doing serious empirical work. In sport we know exactly what the objective is: win the match, win the league. The rules are explicit and enforced. The stakes are high: prize money, contracts, reputations. The participants are experts: professionals who have repeated the same task thousands of times. And most importantly: each action, especially in modern professional sport, is recorded in microscopic detail.
This is not the world most economists live in. In ordinary labour markets, we rarely observe all the inputs and outputs of a production process. Objectives are fuzzy, incentives are opaque, data are noisy. With sport, by contrast, it is as if someone has taken the messy world, put it under bright lights, drawn white lines around the action, and provided a scorecard.
One famous example from this literature comes from penalty kicks. A penalty is a neat two-player, zero-sum game: kicker versus goalkeeper, fixed distance, clear pay-offs. Palacios-Huerta collected more than a thousand penalties and tested a central prediction of game theory: that at equilibrium players should mix their strategies in such a way that they are unpredictable and indifferent between left and right. He finds exactly that. For most professionals, scoring probabilities are equalised across directions, and there is no serial correlation in their choices. In other words, when they stand on the spot, they really do behave as if they had internalised Von Neumann.
Charming as that is, it is when we move to issues like discrimination or incentives that the power of sports data for economics really becomes clear. Think of racial discrimination in labour markets. Gary Becker taught us that in a competitive market, employers who allow prejudice to shape hiring decisions should be punished: if you refuse to hire black workers who are just as productive as white workers, you leave talent on the table and lose out to less prejudiced rivals. Testing that idea properly in most labour markets is very difficult. We do not usually observe the marginal contribution of each worker to firm performance.
English professional football comes close. In the 1970s and 1980s some clubs had far more black players than others. Their wage bills were known; their league positions were known. Early work by Stefan Szymanski showed that, holding the wage bill fixed, teams with more black players tended to end up higher in the table. It looked very much as if some owners and managers still discriminated, and that those who did so paid a competitive price for their prejudice.
The next chapter of this story takes us into the Premier League era. With television money flooding in after the early 1990s, the league became richer, more global and more competitive. Subsequent work surveyed in The Beautiful Dataset shows that as the league became fiercer, the statistical relationship between racial composition and league position disappeared, and the wage premium for white players faded out after the mid-1990s. In a sufficiently competitive environment, in other words, you could no longer afford to indulge your racial tastes. Field a weaker team because of the colour of your players’ skin and you would be punished on the pitch and eventually in the accounts.
That is not just a football story. It is a vivid case study in how competition can discipline discriminatory behaviour, in a way that is very hard to observe in factories or banks or universities, where output is not recorded in a neat league table every year.
What does any of this have to do with South Africa or with African economics more generally? Quite a lot, I would argue. In much of Africa, our complaint as economists is that we lack good data. Labour force surveys are infrequent, firm-level data are thin, tax records are patchy. We often work with small samples and noisy proxies. Yet every weekend the same countries produce extraordinarily rich data on football, rugby, cricket and many other sports, especially since the rise of sports betting. We know the score of every match, the line-up, who scored, who missed, who travelled from where, in what conditions, and often now every pass and tackle as well.
This makes sport a rare exception in data-poor environments. It is the part of the economy where we count obsessively. That creates opportunities. If you want to study how managers respond to pressure and whether mid-season sackings work, there are hundreds of coaching changes in African football that can be analysed. If you wish to understand the costs of long-distance travel, the fixture list of the CAF Champions League offers precisely defined journeys across time zones, climates and altitudes, with performance measured ninety minutes later. If you are interested in the impact of racial quotas or integration policies, South African rugby and cricket have decades of detailed selection and performance records that can be used to study the long and uneven retreat from apartheid.
For economists, especially in Africa, the lesson seems clear. Where the rest of the economy is in the dark, sport is under floodlights. It would be wasteful not to use that light.
Some of this thinking has already gone back into the game itself. Arsenal were early adopters of data by football standards, buying an analytics company more than a decade ago and building an in-house team of data scientists and modellers. Under Mikel Arteta, that tendency has intensified; by all accounts the club leans heavily on data and even artificial intelligence to dissect matches, scout players and plan training. The club of Henry and Tierney now runs on something very close to a beautiful dataset.
As economists, especially those working in and on Africa, we might do worse than to join them, not in designing the next high press at the Emirates, but in using the same wealth of sports data to understand how people respond to incentives when the rules are clear and the score is visible to everyone.
‘How sport can make economics better’ was first published on Our Long Walk. The images were created with Midjourney v7.
Fourie, Johan, and Maria Santana-Gallego. 2017. “The invisible hand of Thierry Henry: How world cup qualification influences host country tourist arrivals.” Journal of Sports Economics 18(7): 750-766.
Palacios-Huerta, Ignacio. 2025. “The Beautiful Dataset.” Journal of Economic Literature, 63 (4): 1363–1423.