Olympic Games: Natural experiments for macroeconomic anticipation effects

Markus Brückner, Evi Pappa

16 September 2011



For years economists have been debating whether the organisation of the modern Olympic Games is a matter of honour or a matter of money, with no clear answer. Indeed, the debate may well be as old as the Olympics itself.1

  • The economic benefits of hosting the Games are dubious for most academics who have conducted independent research on the issue (see for example Owen 2005).
  • Only recently, Rose and Spiegel (2011), using a variety of trade models, show that hosting the Olympics has a positive impact on national exports which is statistically robust, permanent, and large.

These authors find that unsuccessful bids to host the Olympics have a similar positive impact on exports and explain their finding by arguing that what matters is the signal countries send to international markets when these countries bid for the Olympics. However, as the authors also recognise, their explanation cannot fit all aspects of the data. For instance, they cannot explain why open economies should bid to host the Olympics, and why countries bid repeatedly for the organisation of such events.

New research on the economic impact

The starting point of our analysis is that bidding to host mega events such as the Olympic Games creates important demand-anticipation effects that stimulate current output, consumption, and investment (Brückner and Pappa 2011). In other words, bidding for a mega event such as the Olympic Games is “news” about future investment opportunities and future increases in public spending. Thus, private agents may find it optimal to react to the news signal before the event happens.

We view the competition for hosting the Olympic Games as a natural experiment to test for anticipation effects in macroeconomics. This is because the Olympic bidding process begins nine to ten years before the actual hosting of the event. Approximately nine years before the event a technical evaluation of the starting bids by the IOC determines the top five candidates. A host city is selected seven years before the actual organisation of the events and the successful bid delegation signs the "Host City Contract" with the IOC.

Hence, agents in candidate cities/countries receive signals for possible changes in aggregate demand. Eight to nine years before hosting they receive the bidding signal and seven years prior to the actual organisation of the Games they receive the winning or failing signal. The bidding signal is likely to be less informative than the winning one. It gives typically a 1/5 probability to the bidding country for holding the event in eight years, while the winning signal is very informative and produces important news for what is going to happen in the following years. This particular information structure allows us to test whether such news affects economic behaviour and explicitly examine the role of expectations and uncertainty realisation in shaping macroeconomic outcomes.

The research design

To analyse the anticipation effects of Olympic “news” we use a panel of 184 countries spanning the period 1950-2006 and examine how within-country variations in per capita GDP, consumption, investment, government expenditures, the price level, and the exchange rate are related to an Olympic Games indicator variable – both for countries that bid and win the Games and for countries that bid and lose the Games. We employ panel fixed effects estimation techniques that allow for contemporaneous, future, and lagged effects of the Games indicator. We consider the bidding and hosting of the Olympic Games as natural experiments in the sense that candidacies and actual hosting are exogenous to current macroeconomic developments.2

In our study, the Olympic Games turn out to be economically beneficial, but neither for their legacy effects, nor for the “honour” they yield.

  • They are beneficial because of the positive effects they induce on the expectations of private agents about changes in future demand.
  • Hosting the Games generates positive investment, consumption, and output responses before, during, and after celebration.
  • Anticipation effects are present in all the variables we consider; GDP growth increases significantly during the five years before hosting and the peak response occurs four years before the actual hosting of the event.

This increase in GDP per capita growth is due to a significant positive and quantitatively large increase in private investment and consumption. Government spending also increases four years in advance of the actual event. The variable that mostly reflects the anticipatory demand effects of the Games is the price level and the exchange rate. They react significantly when a country bids for the Olympics, reaching their maximum reaction around the time when the announcement of the winner is made.

Figure 1 presents the cumulative effects (and 90% confidence bands) of the Olympic winner´s announcement. The figure is clear about the importance of anticipatory effects in that all variables, apart from government spending, increase significantly prior to hosting – with prices and exchange rates reacting more significantly immediately after the announcement.

Anticipation effects also account for the increases in output growth of unsuccessful bidders. Forward-looking investors should boost investment demand in countries that bid for the Olympics since in those countries expected profits increase. We show that this is the case in the data. We observe significant positive output growth, private investment, and private consumption responses in the bidding countries about seven to ten years before actual hosting. Private investment significantly decreases two and three years after the revelation of the bidding failure indicating that the investment projects undertaken while bidding are mostly reversible and for that reason the after-effects of bidding for the Olympic Games are significantly negative. These results are summarised in Figure 2 where we plot the cumulative multipliers for the host countries (diamonds line) and the bidding countries (circles line) for the variables of interest.

Our results survive a series of sensitivity analyses concerning data treatment, sample periods and omitted variables.

  • First, we compare the organisation of the Olympics with other international events such as the International Expo and the football World Cup.

Both events confirm the presence of anticipation effects, but their effects are not comparable in terms of size with the ones of the Olympics. Countries that have hosted International Expos experienced a significant increase in their real per capita GDP growth before the hosting of the event, but this effect is small and transitory when compared to the one generated by the Olympics. The hosting of the football World Cup, on the other hand, generates negative effects on output growth and only positive effects on government spending growth. Thus, agents perceive the organisation of such an event as a minor investment opportunity which simply leads to increases in government spending crowding out private demand.

  • Second, we investigated whether local effects are stronger than country effects by repeating our analysis using regional data.

The regional analysis confirms the general findings – hosting the Olympics generates positive output growth effects before, on impact, and after the event.

  • Third, stock price data confirm the presence of anticipatory effects.

The stock price index increases significantly eight and nine years before the hosting of the Games and the magnitude of the anticipatory increase in stock prices is comparable in bidding and hosting countries.


Figure 1. Cumulative effects hosting countries

Figure 2. Cumulative effects hosting versus bidding countries



Brückner, M and E Pappa (2011), “For an Olive Wreath? Olympic Games and Anticipation Effects in Macroeconomics”, CEPR Discussion Paper 6844.

Owen, J (2005), “Estimating the Cost and Benefit of Hosting Olympic Games", The Industrial Geographer, 1:1-18.

Rose, A and M Spiegel (2011), “The Olympic effect”, Economic Journal,121:652-677.



1 According to Herodotus, in ancient Greek times when the Olympics were on, Xerxes and Mardonius asked a group of Greek deserters what prize the Olympic winners should get. The answer was “An olive-wreath."   Tigranes, one of Xerxes´ generals, uttered: “Good heavens! Mardonius, what manner of men are these against whom you have brought us to fight – men who contend with one another, not for money, but for honour!”


2 In particular, we use the following econometric specification to estimate the effects of the Games:

   Yi,t = a0Hosti,t + A(L)Hosti,t +B(F)Hosti,t +c0Bidi,t + C(L)Bidi,t +D(F)Bidi,t +αi+βt +ei,t    (1)

where, Hosti,t is an indicator variable that is unity in country i and year t if the country hosted in year t the Olympic Games. Bidi,t is an indicator variable that is unity in country i and year t if the country was bidding to host the Olympic Games that were held in year t. Because bidding to host the Olympic Games takes place about eight to ten years before the actual hosting of the Games, we include up to ten leads of the bidding and hosting country indicator variable on the right-hand side of the estimating equation, so that B(F) = b1F + b2F2 + ... + b10F10 and D(F) = d1F + d2F2+ ... + d10F10. The coefficients in the polynomial B(F) and D(F) capture the anticipatory effects of hosting and bidding for the Olympic Games. Similarly, we examine the delayed-effects of the Olympic Games by including up to ten lags on the right-hand side of the estimating equation, such that A(L) = a1L +a2L2+ ... + a10L10 and C(L) = c1L +c2L2+ ... + c10L10. The contemporaneous effects of the Olympic Games for the bidding and hosting countries are in turn captured by the coefficients a0 and c0. We examine whether bidding and hosting the Olympics has long-run (i.e. permanent) effects on the outcome variables Yi,t by testing whether the sum of the estimated coefficients on the contemporaneous, after and before dummy variables is significantly different from zero.




Topics:  Frontiers of economic research Macroeconomic policy

Tags:  expectations, Olympic Games, sports

Professor of Economics, University of Queensland

Full Professor, European University Institute