As the Football Perspectives readers know, Federation Internationale de Football Association (FIFA) calculates and announces the FIFA world rankings every month. The official name is FIFA/Coca-Cola World Ranking. Male rankings are published monthly, whereas female rankings are published every quarter. The ranking order is based on ranking points that depend on the results of international matches between national teams. These ranking points may reflect the skills of national teams, and the more ranking points held by a national team the happier the citizens of the nation will be, indicating considerable improvement in national welfare. Thus, analysing the determinants of FIFA ranking points is an important economic issue.
Taking each national team’s FIFA world ranking points as a proxy for the proficiency of a nation in international football, I examine the football technology transfer effect and ask whether having more players in top-level leagues improves the national team. I would like to show the main result of my research paper (Miyazaki 2013, available here) as well as literature review.
Other sports may also make people happier, but their popularity varies from country to country. I implicitly assume that football is loved in every country in the world. I hope the Football Perspectives readers will agree with this assumption.
Literature review
To date, several researchers have tackled this problem. Houston and Wilson (2002) used FIFA ranking points as a proxy of the popularity of football. Using the June 1999 FIFA points, they found that football is more popular in richer countries and the marginal popularity decreases with income level. In contrast, Hoffmann et al. (2002) used the January 2001 FIFA points not only to confirm the result of Houston and Wilson (2002), but also to find that a geographical factor is important for determining FIFA points. Specifically, when the average national temperature differs from 14 degrees Celsius, the FIFA ranking points of the corresponding national team decline. Furthermore, Hoffmann et al. (2006) applied a similar analysis to women’s football and found that temperature has an insignificant effect, and that the state system and the gender bias are more crucial. In a socialist country with gender equality the women’s national team is stronger.
The previous studies are based on cross-sectional analyses, which involves data collection at one specific point in time. However, football technology might be affected by other unobservable factors. To remove such factors, Yamamura (2009, 2011) conducted a panel analysis, in which the data are usually collected over time and over the same individuals. (For the panel analysis, please read an undergraduate econometrics textbook such as Stock and Watson, 2010.) Yamamura (2009) used a panel data set of FIFA points over the period 1993–1998 to demonstrate the converging variability of FIFA points, and found that the convergence is explained by skills transfer. That is, players in countries with developing football leagues (not in Europe and Latin America) learn the game in the top professional leagues in England, Germany, Italy, and Spain and then transfer their skills to their domestic national team. Furthermore, Yamamura (2011) used the same data set and found that language differences create communication difficulties when playing football and have a negative effect on football skills.
However, there are two problems with Yamamura (2009). The first is the sample period. FIFA first published a ranking of its member associations in December 1992. Until 1999, a team received one point for a draw or three points for a victory in FIFA-recognised matches, while ignoring important measures including the number of goals and the regional strength. In February 1998, Japan ranked ninth because there are many countries in Asia, and won many preliminary games of the 2000 World Cup played in Japan. To measure the strength of national teams more accurately, FIFA updated the ranking system in January 1999 and readjusted it in June 2006. Thus, using the ranking points from the period 1993–1998 might lead to inaccurate estimates of football skills. The second problem is the proxy of skills transfer. To estimate the effect of skills transfer, Yamamura (2009) utilised the average world ranking points for the most advanced countries having a top league. He insisted that the superior skills in advanced countries make less developed nations catch up with the more advanced ones via skills transfer through international player mobilisation. However, this implies that the average world ranking points for the most advanced countries would have the same effect on all these countries. Rather, the skills transfer effect depends on the number of football players in top-level leagues.
Main results of my paper
I hypothesise that the greater the number of football players in top-level leagues, the more powerful their national team becomes, via skills transfer. Taking each national team’s FIFA world ranking points as a proxy for the proficiency of a nation in international football, I examine this football skills transfer effect. I use panel data from FIFA member nations for 1999–2005 to control for unobserved nation-specific effects. Additionally, allowing for reverse causality, by which players in powerful national teams tend to play in top leagues, I use real purchase power parity (PPP) to conduct an instrumental variable method. Economists often use this method because they deal with observational data involving reverse causality. (For this point, please also read Stock and Watson, 2010.)
I built an econometric model, in which the explained variable is FIFA ranking points (annual average), and the explanatory variables are average local ranking points, average ranking points for the most advanced nations, the year a nation first became a FIFA member, total number of players in the top professional leagues, total number of World Cup appearances, total number of Youth World Cup appearances, real gross domestic production per capita (in US dollars, 2005), population (in thousands), real PPP (in US dollars, 2005). These data are obtained from the FIFA official website, EUFO, RSSSF, and the Penn World Table. I combine them to build a dataset for statistical analysis.
The empirical results are summarised as follows. Although the estimated coefficient is not significant, for the data set containing all nations, the number of top-league players has a small negative effect, while for the data set containing only developing nations (Africa, Asia, North America, and Oceania), it has a small positive effect. If African national teams are excluded, the skills transfer effect in developing nations increases substantially. In particular, if an Asian player plays in a top-level league, the FIFA world ranking points of his national team increase by around 30%, and the estimated coefficient is statistically significant at the 10% level.
Note that the current sample period of the data used in this paper is from 1999 to 2005. In June 2006, FIFA modified the calculation method of FIFA ranking points. At present, there is insufficient macroeconomic country data to examine this later period separately. In future, I would like to use an extended post-June 2006 data set to confirm the results of this paper.