International sports and investor sentiment: do national team matches really affect stock market returns?

Abstract

Ashton et al. (2003), Edmans et al. (2007) and Kaplanski and Levy (2010) document abnormal stock market returns on the trading day following international sporting events, particularly soccer. This study examines returns in matching countries and finds that unusual returns also exist in those countries even though their national teams did not play. The evidence shows that national team matches do not affect neutral markets like the matching countries, which implies that sports do not cause unusual returns in either domestic or foreign markets. The results indicate that changes in investor sentiment following international sports matches do not have a significant effect on asset prices.

Acknowledgements

The author thanks Paul Asquith, Alex Edmans, Diego Garcia and Oyvind Norli for their valuable comments.

Notes

¹ There is a related literature, including Brown and Hartzell (2001) and Bernile and Lyandres (2009), that studies the effect of sports matches on the returns of publicly traded sports teams. For publicly traded teams, wins and losses affect the financial condition of the firm because successful teams are generally more profitable and hence, more valuable. Thus, there is a real economic channel for a team's performance in games to affect its stock price. For national team matches and national stock market indices, which are the subject of this article, there is no similar direct channel for national team performances to affect aggregate stock market indices. Thus, any effect of national team matches on stock market returns is likely to occur through its effect on investor sentiment.

² Markets face down days after World Cup ousters’, Boston Globe, 7 July 2006.

³ For example, Latin American countries, which played in 14.5% of the World Cup matches in the sample, account for less than 3.5% of the market value of the World Market index; so, returns in that region will have little impact on the index.

⁴ For example, Western European countries make up 71.7% of the winners, but only 56.9% of the losers, while Asian and Pacific countries account for 3.1% of the winners and 12.4% of the losers.

⁵ Edmans et al. (2007, p. 1980) recognize this potential problem. To account for it, they form portfolios of winners and losers for each game date and average the residuals on that date for each of the portfolios. However, those portfolios do not adequately account for regional variations in stock market performance because the winner and loser portfolios often include countries from different parts of the world. For example, both the winner and loser portfolios associated with stock market returns on 10 June 2002 include markets from Africa, Asia, Eastern Europe, Latin America and Western Europe.

⁶ If the first-choice country's national team played in the World Cup on the same day and had the opposite result, it is used as the matching country. If the hypothesized effect holds, this rule will make the results stronger because World Cup winners are matched with losers, and vice versa, so that the excess returns of the winner include the effect of the match on the loser, and vice versa.

⁷ There are no data available for countries that border Nigeria and South Africa so those countries are matched with the World Market index.

⁸ The correlations were estimated using data from 1990 to 2002. During the period of 1974 to 2002, correlations rose substantially among markets in the sample. For example, the correlation between stock market index returns for Germany and France from 1974 to 1989 is 0.31 while it is 0.77 for the period 1990 to 2002. Further, many of the countries in the sample do not have data prior to 1990. If all available data are used to estimate the correlation, the maximum-correlation country is almost always a country with data only available in the latter part of the sample period.

⁹ In a few cases in which the total market index is not available, other indices from DataStream are used.

¹⁰ The regions are Western Europe, Eastern Europe, South America, North America, Asia and Africa.

¹¹ The dataset for the other sports was constructed using the countries, time periods and sources described by Edmans et al. (2007). Although the estimates of the loss effect in Table 6 are similar to those in their Table 1, there are some discrepancies in the two samples. For example, they include returns following 9 hockey wins for Germany, while this dataset includes 25. There are smaller discrepancies for several other countries.

¹² The weights are determined by the number of matches, including wins and losses, a national team played during that year's World Cup and the return is the mean return on the national stock market index for all of the trading days during the World Cup. For example, Germany played 5 matches during the 2002 World Cup, out of a total of 328 matches in the sample; so, the weight on Germany's return during that World Cup is 1.52%.

¹³ Following Kaplanski and Levy (2010), all reported results are based on the NYSE Equal-Weighted Index from January 1950 to December 2007. Substituting the NYSE Value-Weighted Index yields very similar results.

¹⁴ World Cup tournaments last about 1 month and shifting the dates by 4 weeks controls for weekday effects. For example, a Monday after weekend World Cup games corresponds to a Monday 4 weeks earlier.

¹⁵ Survey of Current Business, July 2007

¹⁶ ‘Euro 2008 Nets Huge Audience’, Variety, 13 June 2008.

¹⁷ For example, the US Government's Report on Foreign Portfolio Holdings of US Securities shows that the percentage of US equity owned by foreigners increased from 4.7% in 1978 to 11.3% in 2007.