The economic effect of hosting a mega sport event : the effect on unemployment of hosting the FIFA World Cup

(1)

The economic effect of hosting a

mega sport event

The effect on unemployment of hosting the FIFA World Cup

Nick Verkerk 29-06-2016

10359494 University of Amsterdam

Macroeconomics / International economics Ron van Maurik Economics and Finance

Abstract

The major objective of this study is to investigate the effect of hosting the FIFA World Cup on the unemployment rate of a host country. This study investigated the last seven World Cups

over the past 26 years. On the basis of the results of this study, it can be concluded that the unemployment rate of a host country decreases, but only during the year of the World Cup.

(2)

Statement of Originality

This document is written by Student Nick Verkerk who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Contents

1 Introduction………...4

2 Literature review……….5

2.1 What is a mega sport event?...5

2.2 Impact on economic growth………5

2.3 The costs of hosting a mega sport event………7

2.4 The effect on tourism………..8

2.5 The effect on employment……….10

3 Data………13 4 Methodology………..15 5 Results………18 6 Discussion………20 7 Conclusion………22 8 Appendix………23 9 References………29

(4)

1 Introduction

The Football World Cup and the Olympic Games are two of the biggest sport events in the world. Hosting such an event has a huge impact on the economy of a country. Since hosting such an event involves large investments from the public and private sectors, host countries hope to acquire significant revenues and boost their economy. Furthermore, it is often suggested that hosting a mega event will provide job

opportunities, because it requires a lot of preparation. Different studies analysed the employment effect of hosting a mega sport event and the results are inconsistent. For example, Hagn and Maennig (2008) analysed the employment effects of the 1974 Football World Cup and found no significant effect on employment. Another study was conducted by Kasimati and Dawson (2009), who analysed the employment effect of hosting the Olympic Games. They found that unemployment rates

decreased after the announcement that Greece would host the 2004 Olympic Games in 1997. Both these studies were based on one single event in one country. In this paper, the employment effect of seven different World Cups will be analysed by doing a panel analysis and by using dummy variables for the occurrence of the World Cup and the announcement of the new host country of a World Cup. This is the first time that a dummy variable for the occurrence of the announcement of a new host country is used to analyse the employment effect of a World Cup. This study shows that hosting the World Cup significantly decreases unemployment during the year of the World Cup. This paper will discuss several economic effects of hosting a mega sport event and will in particular focus on the employment effect of hosting the World Cup to answer the research question: What is the effect on unemployment of hosting the World Cup? The first section of this paper consists of a review of literature in which some of the economic effects of hosting a mega sport event will be discussed using results of other studies. The second section of this paper will elaborate on the methodology of this study. The third section will present the results of this study, the fourth section discusses the results and the last section draws a conclusion on both the results of existing studies and the results of this study.

(5)

2 Literature review

Hosting a mega sport event costs a lot of money, but how big are the revenues and where do they expect to make money? According to economists, organizing a mega sport event can boost the economy of a country. The main source of revenue comes from tourism, but hosting such an event also leads to investments in infrastructure and an increase in consumption which leads to extra growth opportunities (Sterken, 2006, p.388). According to Gratton, Shibli and Coleman (2006), the World Cup has the potential to generate a large economic impact for host nations. Several studies have assessed the impact of hosting a mega sport event and this section will discuss some economic effects of hosting a mega sport event.

2.1 What is a mega sport event?

The FIFA World Cup and the Olympic Games can be considered as a mega sport event. The distinction between a sport event and a mega sport event is based on the size (Müller, 2015, 628). Müller created a scoring matrix in which four dimensions can be classified in different ranges. Every range has a different score. When the points are summed up, the results show whether the sport event can be classified as a major sport event, a mega sport event or a giga sport event. The four dimensions Müller takes into account are: visitor attractiveness, mediated reach, cost and transformative impact. Visitor attractiveness is the actual amount of spectators who visited the sport event. The mediated reach are spectators that did not visit the event itself, but stayed home and enjoyed the event through the media. By cost, Müller means the cost of hosting the sport event and with transformative impact he accounts for the amount of invested capital of the host country. According to Müllers scoring matrix, the FIFA World Cup and the Olympic Games can be considered as mega sport events.

2.2 Impact on economic growth

Most of economists agree on the idea that hosting a mega sport event has a very small or no impact on economic growth. On the short term, there could be an increase in economic growth due to the rise in consumption, investments and production, but on the long-term the economy converges again.

For instance, Baade and Matheson (2004) did an ex post analysis on the 1994 World Cup held in the United States. They selected explanatory variables from

(6)

existing models to predict economic activity in the absence of the World Cup. They estimated the economic impact by comparing the projected level of economic activity without the event to the actual levels of economic activity that occurred in cities hosting matches. Their analysis suggests a 93,60% chance that hosting the World Cup has a negative economic impact on the host communities, because it requires substantial expenditures on infrastructure and security. Loots (2006) performed a study on the same World Cup in the United States. He found that nine of the host cities experienced income losses and only four experienced income gains.

Furthermore, his study showed that non-host cities experienced higher income growth than host-cities, as non-football-related tourism was diverted from congested host cities to non-host-cities (Saayman & Rossouw, 2008, p.3)

Szymanski (2002) analysed the economic impact of hosting a major event, such as the World Cup or the Olympic Games by analysing the economic growth of a country. Szymanski (2002) used data on the twenty largest economies measured by current GDP over the last thirty years. Most of these countries hosted at least one of those events. Table 1 (Szymanski, 2002, p.176) shows the impact on economic growth of hosting a mega sport event.

(Source: Szymanski, 2002, p.176)

The impact on economic growth can be seen as statistically significant if the t statistic is larger than 1.96 in absolute value (significance level of 5%). Table 1 shows that the only statistically significant economic impact takes place during the year of the event. This t-statistic is negative, suggesting that economic growth declines rather than increases (Szymanski, 2002, p.176).

Another study focussing on the effect of hosting a mega sport event on economic growth was conducted by Sterken (2006). Sterken analysed the growth

(7)

impact of mega sport events by presenting ex-post-cross-country event results for the Olympic Games and the FIFA World Cup. Sterken (2006) used data from the World Development Indicators from the World Bank. This set is available for 208 countries from 1960 onwards, and allowed him to get consistent data on GDP per capita, gross fixed capital formation and trade data. He found that there is a positive effect on GDP of hosting the Olympic Games and a slightly negative impact on GDP of hosting the World Cup. One reason for this difference is that the International Olympic

Committee selects higher growth potential economies to organize the Olympic

Games than FIFA does for the World Cup. A second reason might be that hosting the Olympic Games requires more investments, because of the larger variety of sports (Sterken, 2006, p.378-388).

All above studies show that hosting the World Cup does not have a positive effect on economic growth. These results show that the effect is mostly insignificant, but the few significant results show a negative effect on economic growth. Although there are expected negative effects of hosting a mega sport event, countries still aim to host these events because of the expected benefits. A successful event will create positive spin-offs due to international publicity and recognition. This causes potential host countries to ignore the negative economic impact that might occur (Kim et al, 2006, pp. 88-89).

2.3 The costs of hosting a mega sport event

Hosting a mega sport event costs a lot of money in the short run, because it requires substantial expenditures on infrastructure. In the case of the World Cup, the FIFA requires that the host country provides at least eight and preferably ten modern stadiums with a capacity of 40,000 to 60,000 spectators. For the World Cup hosted by Japan and South Korea in 2002, each of the 2 host countries had to provide ten separate stadiums. Because both countries did not have the sporting infrastructure to host such an event, South Korea had to build ten new stadiums at a cost of nearly $2 billion and Japan had to build seven new stadiums and refurbished three others at a total cost of at least $4 billion (Baade & Matheson, 2004, p. 345). Some of these costs should not be equated with costs of the World Cup. If the stadiums remain in use after the World Cup, the World Cup related costs for stadiums should not be the cost of the stadium itself, but the costs in the form of losses in the value of the stadiums due to the tournament. Furthermore, it should be noted that new or

(8)

renovated stadiums engender a novelty effect: curiosity, increases in comfort,

improved views and a better atmosphere lead to significantly higher spectator figures, at least for a period after the improvements (Allmers & Maennig, 2009, p. 509). Hosting a mega sport event is not only very costly in the short run, but also affects the long run. During the event there are operational costs, such as keeping the event safe and after the event there will be maintenance costs (Sterken, 2006, p.380). Sometimes these maintenance costs are too high and lead to deserting their recently build facilities because of the high costs. One example is the Olympic Games in Sochi. The costs of hosting the Olympic Games in Sochi were 55 billion dollars, which makes it the most expensive Olympic games ever (Müller, 2014, p.628). The high costs would not be so problematic, if they indeed led to significant revenues and the Olympic Park would still be of use after the Olympic Games ended (Müller, 2014, p.639), however after the Olympic Games, the Olympic Park in Sochi is barely used at all. The Park sees only few tourists because of the absence of attractions and its remoteness relative to the city centre and the main beaches. The Olympic Stadium, which was used for the opening and closing ceremony, is now under reconstruction and will be used for the 2018 World Cup in Russia. The

reconstruction costs an additional 52 million dollars. What will happen after the four to five matches of the World Cup have been played is unclear since Sochi lacks a

football club to fill a stadium of this size (Müller, 2014, p.643).

With regard to the high costs of a mega sport event, Deccio and Baloglu (2002) state in their study that mega sport events are likely to cause price inflation and an increase in local tax to construct the facilities that are required to host the event.

2.4 The effect on tourism

The tourism effect is one among several that host cities seek, arguing that the

international media coverage preceding and during the a mega sport event presents a tremendous opportunity to advertise themselves in the global marketplace

(Kasimati, 2003, p.438). Also, tourism is the main resource of revenue when hosting a mega sport event. It is reasonable to say that hosting a mega sport event draws significant numbers of domestic and international tourists.

Most ex ante studies expect the tourism sector to be one of the main

(9)

For instance, a total of 309,554 foreign tourists arrived in South Africa for the primary purpose of attending the 2010 FIFA World Cup (Fifa.com, 2010). On the other hand, tourists who are less World Cup-enthusiastic might postpone a planned trip to the host nation or even cancel it on account of the event (Allmers & Maennig, 2009, p. 506). To analyse the effect of hosting the World Cup, Allmers and Maennig (2009) directly studied the number of tourists’ overnight stays, national income from tourism and the retail sales for the 1998 World Cup in France and the 2006 World Cup in Germany. The results of their study show that the number of tourists’ overnight stays increased significantly in Germany in 2006. The regression, however, did not indicate any significant effects on the overnight stays in France. This study also illustrates a rise in income from international tourism during the World Cup in 2006 in Germany since these results are significantly different from zero on the 1 percent error level. As mentioned above, France did not register an increase in the number of tourists’

overnight stays in 1998. By the same token, France did not register any significant increases in income from international tourism during the World Cup in 1998. The retail industry usually hopes for positive effects from hosting a WC due to increased foreign and domestic consumption (Allmers & Maennig, 2009, p.507). For both the 1998 World Cup in France and the 2006 World Cup in Germany, no significant results were found on the effect of hosting the World Cup on retail sales.

Another study, which analysed the effect on tourism of the 1988 Summer Olympic Games in Korea, found significant increases in tourism during the event as well as after the event. Remarkably, the greatest increase in tourism is observed the year following the prestigious event. However, the researchers find that the increase in tourism is not permanent and thus diminishes over time (Kang & Perdue, 1994, p.205).

As mentioned above, not all income from tourism is World Cup related. Lee and Taylor (2005) share this opinion and they also state that World Cup tourists spend 1,8 times more than foreign leisure tourists (Lee & Taylor, 2005, p.595). In their paper they split up tourists in two groups: World Cup visitors and Ordinary tourists. Their paper reports on the assessment of the 2002 World Cup in South Korea and Japan, using an estimation method that excluded tourists whose travel was not World Cup related. In this way, they directly attribute tourist numbers and expenditures to the event. This study shows that 57.7% of the tourists have especially travelled to South Korea because of the World Cup. For this study, the

(10)

economic impacts were estimated by using input-output analysis and multiplying the direct World Cup tourist expenditure by tourism sectoral multipliers for output,

income, employment, value added, indirect tax, and import. For instance, the tourism output multiplier for the restaurant sector of 2.86 indicated that every dollar spent by World Cup tourists generated US$2.86 of output for the Korean economy. The results of this analysis indicate that World Cup tourists’ expenditure of 522 million dollars generated 1.35 billion dollars of output as a result of the multipliers.

One other factor that influences the magnitude of the increasing tourism is whether the sport event is held in the off-season or in the peak-season. Fourie & Santa-Gallego (2011) show that sport events held in the off-season show a higher increase in tourism than was predicted, while events held in the peak-season show a decline in predicted tourism. One explanation for this is mentioned before: tourists who are less World Cup-enthusiastic might postpone a planned trip to the host nation or even cancel it on account of the event (Allmers & Maennig, 2009, p. 506).

2.5 The effect on employment

Hagn and Maennig (2008) examined the employment effects of the 1974 Football World Cup on the 9 host cities (Berlin, Dortmund, Düsseldorf, Frankfurt am Main, Gelsenkirchen, Hamburg, Hannover, Munich and Stuttgart). The maximum period of observation stretches from 1961 to 1988.The following graph shows the progression of average employment in the host cities with employment in the other cities over the observation period is showed.

Figure 1: Average employment in WC 1994 cities and non WC 1994 cities

(11)

Figure 1 (Hagn & Maennig, 2008, p.1065) shows that the two lines progressed equally for the most part until 1970. From that point, average employment in the host cities decreased and dropped below the average employment of the non-host cities. From 1976 onwards the average employment in host cities and non-host cities progressed equally again. Hagn and Maennig (2008) analysed whether the

differences in employment development in the two groups are significantly related to hosting of the 1974 World Cup. In order to do so, they used four different estimation approaches and they concluded that the 1974 World Cup in Germany did not

generate a significant employment effect.

The study of Hahn and Maennig found no significant positive effect on employment of the 1974 World Cup, but there are studies that show that hosting mega sport events does have a positive effect on employment. Previous studies found that the Barcelona Olympic Games in 1992 created 20.000 jobs, the Sydney Olympic Games in 2000 created 16.500 jobs and the 2006 World Cup in Germany created 50.000 jobs. However, for the largest part these are temporary jobs, because it is a once-off event that takes place over a short period of time (Saayman &

Rossouw, 2008, p.4).

Kasimati and Dawson (2009) analysed the employment effect of hosting the Olympic Games. Their paper examines the impact of the Athens 2004 Olympic Games on the Greek economy. Annual time series data from 1958 to 2005 were used for the analysis. Their model included GDP as explanatory variable, since the unemployment equation can be derived by using Okun’s law. According to Okun’s law, there is a negative relationship between unemployment and GDP. The results of their study are shown in Figure 2 (Kasimati & Dawson, 2009, p.144). For the period 1997-2005 unemployment fell by 1.9% per year (Kasimati & Dawson, 2009, p. 145). On the long term the effect on employment is modest.

(12)

Figure 2: Unemployment (UN) with and without Olympic Games

(Source: Kasimati & Dawson, 2009, p.144)

Figure 2 (Kasimati & Dawson, 2009, p.144) shows that Greece's level of unemployment decreased significantly as a result of hosting the Games throughout the period 1997–2004. However, by 2005 the two lines appear to converge. The announcement of the nomination of Athens to host the Games was on 5th September 1997, so it is possible that unemployment decreases after the

announcement of a new host country. However, Késenne (2006) states that during a mega sport event a crowdingout effect occurs. When the host country starts

preparing for the event, the new labour needed for the extra construction activities will be pulled out of other construction activities. As a result, this may not lead to a decrease in unemployment, because no new construction workers are hired.

Baade and Matheson (2002) did an ex-post analysis of the Summer Olympics of 1984 and 1996. Their approach was based on the estimation of the level of

employment in the Olympic Games’ absence. They constructed an econometric equation including population, income, wages and taxes as independent variables and used dummy variables for the occurrence of the Olympics and the oil boom. They found no significant results.

Malfas, Theodoraki and Houlihan (2003) conducted an ex-post analysis on the effect on unemployment and found that a mega sport event creates a significant

(13)

amount of job opportunities. They state that the job opportunities are not the direct results of the mega sport event but could alternatively result from the tourism sector. More employees are needed due to the increase in visitors. Their study indicates that, for the 1992 Olympic Games in Barcelona, the unemployment rate decreased by 8.8% (Malfas, Theodoraki & Houlihan, 2003).

Above studies show that the employment effect of hosting a mega sport event differs. The study of Hahn and Maennig (2009) found no significant positive effect on employment of the 1974 World Cup, whereas the study of Baade and Matheson (2002) showed no significant effect for both the 1984 Olympic Games and the 1996 Olympic Games. The studies of Kasimati & Dawson (2009) and Malfas, Theodoraki and Houlihan (2003) showed a significant decrease in unemployment. Kasimati and Dawson found a significant decrease in the unemployment rate after the

announcement of a new host country was made and Malfas, Theodoraki and Houlihan found a significant decrease in the unemployment rate during the event.

3 Data

For this study, relevant annual data from the Worldbank from 1980 until 2014 on the unemployment rate, government expenditure, consumption, national income, GDP, and population for the last eight host countries are used. All the variables, except from the unemployment rate, are annual growth percentages. Data are gathered for the last eight host countries: Italy, United States, France, South Korea, Japan, Germany, South Africa and Brazil.

This study focuses on the effect on unemployment, thus the unemployment rate is used as the dependent variable. The unemployment rate is a percentage of the total labour force divided by the share of labour force who is without a job and is seeking for a job. I used population and income as independent variables as they were used in the studies of Baade and Matheson (2002) and Hagn and Maennig (2008) to explain the effect on employment. Furthermore, I also included government expenditure, consumption and GDP since Kasimati (2003) used these variables to explain the effect on employment.

I started the analysis by presenting graphs of the unemployment rates of all the countries separately. These figures show whether the unemployment rate changed due to the hosting of the World Cup. The graphs are shown below.

(14)

0 5 10 15 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

Figure 4: Unemployment

rate USA(%)

0 5 10 15 20 25 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

Figure 3: Unemployment

rate Italy(%)

0 2 4 6 8 10 12 14 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

Figure 5: Unemployment

rate France(%)

0 2 4 6 8 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

Figure 6: Unemployment

rate South Korea(%)

0 2 4 6 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

Figure 7: Unemployment

rate Japan(%)

0 5 10 15 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

Figure 8: Unemployment

rate Germany(%)

0 5 10 15 20 25 30 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

Figure 9: Unemployment

rate South Africa(%)

0 2 4 6 8 10 12 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

Figure 10: Unemployment

rate Brazil(%)

(15)

Firstly, I analysed 1990 World Cup in Italy. In 1984 the announcement was made that Italy would be the host country of the 1990 World Cup. Figure 3 shows small fluctuations in the unemployment rate in the period 1984-1990. This figure suggests that there is a small decrease in unemployment at the year 1990 in which the World Cup took place. The second World Cup in my analysis was the 1994 World Cup in the United States. The announcement was made in 1988. For the period 1988-1994, figure 4 shows an increase in the unemployment rate followed by a decrease in the unemployment rate. It seems like the effect of hosting the World Cup in the United States was ambiguous. The third World Cup that I considered in this study is the 1998 World Cup in France, which was announced in 1992. Figure 5 indicates a big increase in unemployment after this announcement. After the World Cup in 1998, the unemployment rate decreased again. The fourth World Cup is the 2002 World Cup hosted by South Korea and Japan, which was announced in 1998. Figure 6 shows a huge decrease in the unemployment rate in South Korea after the announcement was made. Figure 7 shows a small increase in the unemployment rate in Japan after the announcement was made. One explanation for the huge decrease in unemployment in South Korea is that South Korea had to build ten new stadiums, as mentioned before. The 2006 World Cup in Germany is the fifth WC of this study. This World Cup was announced in 2000. Figure 8 shows an increase in the unemployment rate followed by a decrease in the unemployment rate in the period 2000-2006. The sixth World cup of this study is the 2010 World Cup in South Africa. This announcement was made in 2004. Figure 9 shows a decrease in the unemployment rate after the announcement was made. One explanation is that, similar to South Korea, South Africa required a lot of preparation before hosting the World Cup. The final World Cup of this study is the 2014 World Cup in Brazil, which was announced in 2007. Figure 10 shows an overall decrease in the unemployment rate in Brazil after the announcement was made.

4 Methodology

This paper analyses the effect of organising the World Cup on employment of a country. I started the analysis with graphs of the unemployment rates of all the individual countries. These figures indicate whether the unemployment rate changed as a result of hosting the World Cup. In order to test if these changes are significant, I conducted a panel analysis. I gathered data on the last seven World Cups and made

(16)

a regression model with the unemployment rate as the dependent variable. Because I wanted the change in unemployment rate as a dependent variable I generated a new variable called UNE. I did this by using the command ‘generate UNE =

Unemploymentrate - L.Unemploymentrate’ in Stata. L.unemploymentrate is the lagged variable of unemployment rate. The variable UNE is used because the expectation is that the unemployment rate of year t is heavily correlated with the unemployment rate of year t-1. If the variable unemployment rate was used, there would potentially be omitted variable bias. By using the variable UNE the change in the unemployment rate is used as the dependent variable. I used dummy variables for the occurrence of the World Cup and for the occurrence of the announcement of a new host country to see if there is a significant effect on the unemployment rate of a country. The value of the dummy variable WC is one in the year of a World Cup and is zero otherwise. The dummy variable AnnWC is a trend, because the expectation is that the unemployment rate will decrease more when we get closer to the year of the World Cup. The values of the AnnWC dummy variable can be found below.

The regression model:

𝛥𝑈𝑁𝐸𝑖𝑡 = 𝛽0+ 𝛽1𝑃𝑜𝑝𝑖𝑡+ 𝛽2𝐺𝐷𝑃𝑖𝑡 + 𝛽3𝐼𝑛𝑖𝑡+ 𝛽4𝐶𝑜𝑛𝑖𝑡 + 𝛽5𝐼𝑛𝑣𝑖𝑡+ 𝛽6𝑊𝐶 + 𝛽7𝐴𝑛𝑛𝑊𝐶 +

𝜀

In which:

ΔUNE = Unemployment rate – L. Unemployment rate Pop = Population growth (annual %)

GDP = GDP growth (annual %)

In = Adjusted net national income (annual % growth)

Con = Household final consumption expenditure, etc. (annual % growth) Inv = General government final consumption expenditure (annual % growth) WC= 1 when country hosts World Cup, 0 otherwise

AnnWC= value 1 at t-6, 2 at t-5, 3 at t-4, 4 at t-3, 5 at t-2 and 6 at t-1. In the case of brazil the value is 1 at t-7 and 7 at t-1

(17)

This method is sufficient because this study focusses on different countries and different years. Panel data refers to data containing time series observations of a number of individuals, or in this case countries. Observations in panel data involve at least two dimensions; a cross-sectional dimension, indicated by subscript i, and a time series dimension, indicated by subscript t (HSIAO, 2005, p.144). In this analysis, subscript i indicates the host country and subscript t indicates the year.

Before using this method, I tested a few criteria. I first tested if all the variables were stationary. Most statistical forecasting methods are based on the assumption that the variables are stationary. If the variables are stationary, predictions will be more accurate. In order to test if all the variables were stationary, I used the Levin-Lin-Chu unit-root test in Stata for all variables. Because the Levin-Lin-Chiu test requires strongly balanced data, Germany and South Africa were excluded from the sample due to incomplete data on unemployment rates. For all the other variables all countries were included. The results show that all variables are stationary (see appendix).

The next step was to test whether all the required conditions for the error variable hold. The first condition is that the probability distribution of the error variable is normal. To test this I used the skewness/kurtosis test for normality. I first had to predict the residuals by using the command ‘predict resi, residuals’ in Stata. To use the skewness/kurtosis test for the error variable I used the ‘sktest resi’ command in Stata. The test showed that the probability distribution of the error variable is normal (see appendix).

The second condition is that the mean of the error variable is 0. To test this I used the ‘mean resi’ command in Stata. Results show that the mean of the error variable is 0 (see appendix).

The third condition is that the variance of the error term is required to be constant. When this requirement is violated, the condition is called heteroskedastic (Keller, 2009, p.655). To test whether the condition is heteroskedastic, I used the ‘esstat hettest’ command in Stata. The results show that the variance of the error term is non-constant, so robust standard errors are used in the regression model (see appendix).

The fourth condition is that the values of the error variable should be

independent. Error terms that are correlated over time are said to be autocorrelated (Keller, 2009, p.656). To test for autocorrelation I used the ‘Wooldridge test for

(18)

autocorrelation in panel data’. The results show that the values of the error variable are independent (see appendix).

Variables correlated with the error term are called endogenous variables. I used the Hausman test for endogeneity to determine whether or not there is some form of omitted variable bias in the model. To test this I included the residuals of the variables in the regression. The lowest p-value of this test was the residual of

population and that was a p-value of 0,087. Because no p-value was below 0,05, I did not have to run an instrumental variables regression. The results of the Hausman tests indicate no endogenous variables.

As shown in figure 3-10, the unemployment rate of the countries differ. For instance, the unemployment rate of Italy is about three times as high as the

unemployment rate in South Korea. That is why I included country dummies. I also included time dummies, because this controls for omitted variables that differ

between cases but are constant over time. I used the command xtset Country Year in STATA to include time and country dummies. To control for omitted variables, I tested whether I should use a fixed effects model or a random effects model. I did this by running a Hausman test. The Hausman test checks a more efficient model against a less efficient but consistent model to make sure that the more efficient model also gives consistent results. The Hausman tests the null hypothesis that the coefficients estimated by the efficient random effects estimator are the same as the ones estimated by the consistent fixed effects estimator. The p-value of the test was 0,1132 and therefore was insignificant (see appendix). Following the result of the Hausman test, the random effects model is used in the panel analysis.

5 Results

In this section the results on the empirical research will be presented. Table 3 presents the results of the regression with UNE as the dependent variable and population growth, GDP growth, income, consumption, investment and the dummy variables WK and AnnWC as independent variables.

(19)

Table 3: Regression results

UNE Population growth 0.240 (1.54) GDP growth -0.112* (1.92) Income -0.031 (1.54) Consumption 0.002 (0.10) Investment 0.024 (0.78) WK -0.605* (1.87) AnnWC 0.041 (1.08) Observations 204 R-squared 0.19

Robust t statistics in parentheses

* significant at 10%; ** significant at 5% ; *** significant at 1%

(20)

The results indicate positive coefficients for the variables population growth, consumption, investment and AnnWC. The results indicate negative coefficients for the variables GDP growth and WK. The variable WK, which equals 1 in the year of the World Cup, has a coefficient equal to -0.605 and is significant at a 10% level. The R-squared is quite low, approximately equal to 0.19. This means that 19% of the total variation in UNE is explained by the seven independent variables.

6 Discussion

This section will discuss the results of the empirical research. Based on the model that was presented previously, I will interpret the results considering the effect of hosting the World Cup on employment.

I have conducted a panel analysis in order to find a significant effect of hosting the World Cup on the unemployment rate of a country. Earlier studies on the same topic failed to find significant results. For instance, the study of Hahn and Maennig (2009) show no significant effect on unemployment of the 1974 World Cup in Germany. However, there are studies that found a significant effect on

unemployment as a result of hosting the Olympic Games. One example is the study of Kasimati and Dawson(2009). The biggest difference between this study and the studies of Hahn and Maennig (2009) and Kasimati and Dawson (2009) is that this study does not focus on one single event. This study analysed the last seven World Cups in eight different host countries, so it might be easier finding significant results by using data from different countries and a longer time period.

The regression results indicate a negative effect of the WK variable on the unemployment rate, which is significant at a 10% level. According to these results, the unemployment rate in a host country will decrease during the year of the World Cup. The variable of the announcement of the new host country is not significant. According to these results, the unemployment of the host country will not change from the moment that the new host country is announced until the year of the World Cup itself.

The significant result of the effect on unemployment is in contrast with the study of Hagn and Maennig (2008), who found no significant effect on

unemployment. One reason for this different result is that their study focusses primarily on the 1974 World Cup and this study focusses on the last seven World Cups. However, this result can be traced back to the study of Malfas, Theodoraki and

(21)

Houlihan, who show a significant decrease in the unemployment rate during the event. The study of Kasimati and Dawson (2009) indicates a negative result on unemployment when the announcement of the new host country was made. These results are in contrast with my study, because the effect of the announcement of a new host country had no significant effect on the unemployment rate in my study. One explanation is that there is a difference in preparation between a World Cup and the Olympic Games. It is possible that hosting the Olympic Games requires more preparation and will lead to a decrease in unemployment. One similarity between the existing literature and this study is that the effect on unemployment of hosting a World Cup is modest.

As mentioned before, hosting a World Cup requires a lot of preparation.

Stadiums and other infrastructure have to be build, which requires employees. These results show that the unemployment rate does not decrease until the year of the World Cup, possibly because of the “crowding out effect”. Kénesse (2006) stated that the new labour needed for the extra construction activities will be pulled out of other construction activities.

The decrease in unemployment during the World Cup can be explained by the rise in the tourism sector. This can be traced back to the study of Malfas, Theodoraki and Houlihan (2003), who found a significant decrease in the unemployment rate during the 1996 Olympic Games in Barcelona due to the tourism sector. Also, similar studies of the existing literature found significant increases in tourism during the event and firms need to answer the expansion of demand in the accommodation and food services by increasing their production and therefore hire more employees. Another reason for the decrease in unemployment during the World Cup is the need for extra security. The host country is responsible to employ trained persons to fulfil the duties required for a safe event. This includes national security officers,

stadium/venue security officers, stadium stewards and private security (Fifa.com, 2016).

One way to improve this study is to examine the effect on unemployment in different sectors like accommodation, entertainment and food services. It would be interesting to distinguish between different sectors and to examine which sectors are influenced the most. Another way to improve this study is to use data on the Olympic Games as well, since this study focussed primarily on the World Cup.

(22)

7 Conclusion

This paper analysed the effects of hosting a World Cup on unemployment. Earlier studies on the same topic failed to find significant results. The study of Hahn and Maennig (2009) showed no significant effect on unemployment of the 1974 World Cup in Germany. However, there are studies that did find a significant effect on unemployment of hosting the Olympic Games. One example is the study of Kasimati and Dawson (2009). I used graphs of the unemployment rates of all the host

countries to identify possible effects and then I conducted a panel analysis to try and find significant results.

The graphs show no clear effect on unemployment rates of hosting the World Cup. Therefore, no conclusion can be drawn by using these graphs. The results of the panel analysis show a negative effect of the WK variable on the unemployment rate. According to this result, the unemployment rate in a host country will decrease in the year of the World Cup. Furthermore, the results show that the unemployment rate of the host country will not change from the moment when the new host country is announced until the year of the World Cup. One explanation for this result may be the presence of the crowding out effect. Possible ways to improve this study would be to analyse different sectors of employment and use data of the Olympic Games as well. To conclude and to answer the research question, according to this paper, hosting a World Cup decreases the unemployment rate of the host country, but only during the year of the World Cup.

(23)

8 Appendix

Tests for stationarity of the variables

Levin-Lin-Chu unit-root test for Investment

Ho: Panels contain unit roots Number of panels = 6 Ha: Panels are stationary Number of periods = 35

AR parameter: Common Asymptotics: N/T -> 0 Panel means: Included

Time trend: Not included

ADF regressions: 1 lag

LR variance: Bartlett kernel, 10.00 lags average (chosen by LLC)

Statistic p-value

Unadjusted t -6.4681

Adjusted t* -3.3857 0.0004

Levin-Lin-Chu unit-root test for Consumption

Statistic p-value

(24)

Levin-Lin-Chu unit-root test for Income

Statistic p-value

Adjusted t* -3.3923 0.0003

Levin-Lin-Chu unit-root test for GDPgrowth

Statistic p-value

(25)

Levin-Lin-Chu unit-root test for Populationgrowth

Statistic p-value

Adjusted t* -2.5671 0.0051

Levin-Lin-Chu unit-root test for Unemploymentrate

Statistic p-value

(26)

Levin-Lin-Chu unit-root test for UNE

Statistic p-value

Adjusted t* -6.6695 0.0000

Test for normality

Skewness/Kurtosis tests for Normality Joint

Variable | Obs Pr(Skewness) Pr(Kurtosis) adj chi2(2) Prob>chi2 resi | 204 0.0000 0.0000 . 0.0000

Mean of the error variable

Mean estimation Number of obs = 204

| Mean Std. Err. [95% Conf. Interval] resi | -1.02e-09 .071574 -.1411238 .1411238

(27)

Test for heteroscedasticity

Breusch-Pagan / Cook-Weisberg test for heteroskedasticity

Ho: Constant variance

Variables: fitted values of UNE

chi2(1) = 7.41 Prob > chi2 = 0.0065

Test for autocorrelation

Wooldridge test for autocorrelation in panel data H0: no first order autocorrelation

F( 1, 5) = 0.268 Prob > F = 0.6266

Year of World Cup and year of announcement per country

Country Year of World Cup Year of announcement

Italy 1990 1984 United States 1994 1988 France 1998 1992 South Korea 2002 1996 Japan 2002 1996 Germany 2006 2000 South Africa 2010 2004 Brazil 2014 2007

(28)

Hausman test Coefficients (b) fixed (B) random (b-B) Difference Sgrt(diag(V_b-V_B)) S.E. Population~h .4778721 .2395767 .2382954 .2214895 GDPgrowth -.2028867 -.1116836 -.0912031 .0256026 Income .0102665 -.0309778 .0412444 .0102159 Consumption -.0002871 .0020398 -.0023269 . Investment .0102289 .0235793 -.0133504 .0033413 WK -.5977332 -.6048823 .0071491 . Aanko~Ktrend .0550409 .04101 .0140308 .0052706

b = consistent under Ho and Ha; obtained from xtreg

B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic

chi2(7) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 11.64 Prob>chi2 = 0.1132

Summary statistics

Variable Obs Mean Std.Dev Min Max

UNE 204 -.0063726 1.110278 -10.1 4.4 Population~h 210 .734416 .5509348 -.2003206 2.36681 GDPgrowth 210 2.798687 3.081208 -5.713898 12.26601 Income 210 2.704912 3.912147 -16.44297 16.58711 Consumption 210 2.802192 3.272107 -10.73002 13.45043 Investment 210 2.604491 3.771073 -10.75291 27.17575 WK 210 . 0285714 .1669967 0 1 Aanko~Ktrend 210 .6333333 1.563285 1 7

(29)

9 References

Allmers, S. & Maennig, W. (2009). Economic impacts of the FIFA Soccer World Cups in France 1998, Germany 2006, and outlook for South Africa 2010. Eastern Economic Journal, 35(4), 500-519.

Baade, R., & Matheson, V. (2004). The Quest for the Cup: Assessing the Economic Impact of the World Cup. Regional Studies, 38(4), 343-354.

Baade, R., & Matheson, V. (2002). Bidding for the Olympics: Fool’s gold, Transatlantic sport: The comparative economics of North American and European sports, 127.

Deccio, C., & Baloglu, S. (2002). Nonhost community resident reactions to the 2002 Winter Olympics: the spillover impacts. Journal of Travel Research, 41(1), 46–56

Fedderson, A. & Maennig, W. (2012). Mega-events and sectoral employment: The case of the 1996 Olympic Games. Contemporary Economic Policy, 31(3), 580-603.

Fourie, J., & Santana-Gallego, M. (2011). The impact of mega-sport events on tourist arrivals. Tourism Management , 1364-1370.

Gratton, C., Shibli, S., & Coleman, R. (2006). The economic impact of major sports events: a review of ten events in the UK. The Sociological Review, 54, 41-58.

Hagn, F. & Maennig, W. (2008). Employment effects of the Football World Cup 1974 in Germany. Labour Economics, 15(5), 1062-1075.

HSIAO, C. (2005). WHY PANEL DATA?. Singapore Econ. Rev., 50(02), 143-154.

Kang, Y., & Perdue, R. (1994). Long-term impact of a mega-event on international tourism to the host-country: a conceptual model and the case of the 1988 Seoul Olympics. Journal of International Consumer Marketing , 205-225.

Kasimati, E. (2003). Economic aspects and the Summer Olympics: a review of related research. International Journal Of Tourism Research, 5(6), 433-444.

Kasimati, E. & Dawson, P. (2009). Assessing the impact of the 2004 Olympic Games on the Greek economy: A small macroeconometric model. Economic Modelling, 26(1), 139-146. Keller, G. (2009). Managerial statistics. Australia: South-Western /Cengage Learning.

Késenne, S. (2005). Do We Need an Economic Impact Study or a Cost Benefit Analysis of a Sports Event? European Sport Management Quarterly, 5(2), 133-142.

(30)

Kim, H., Gursoy, D., & Lee, S. (2006). The impact of the 2002 World Cup on South Korea: comparisons of pre- and post-games. Tourism Management, 27(1), 86-96.

Lee, C. & Taylor, T. (2005). Critical reflections on the economic impact assessment of a mega-event: the case of 2002 FIFA World Cup. Tourism Management, 26(4), 595-603. Loots E. (2006). The 2010 Soccer World Cup: An economic and socio-economic perspective. DiscourseDiskoers, 34(2).

Malfas, M., Theodoraki, E., Houlihan, B. (2004). Impacts of the olympic games as megaevents, Municipal Engineer 157(3): 209–220.

Müller, M. (2014). After Sochi 2014: costs and impacts of Russia’s Olympic Games. Eurasian Geography And Economics, 55(6), 628-655.

Müller, M. (2015). What makes an event a mega-event? Definitions and sizes. Leisure Studies , 1-16.

Saayman, M., & Rossouw, R. (2008). The economic value of the 2010 Soccer World Cup. Acta Commercii, 8(1), 1-14.

Stadium Safety and Security - FIFA.com. (2016). FIFA.com. Retrieved 7 June 2016, from http://www.fifa.com/governance/security/stadium-safety-and-security.html

Sterken, E. (2006). Growth Impact of Major Sporting Events. European Sport Management Quarterly, 6(4), 375-389.

Study reveals tourism impact in South Africa. (2010). FIFA.com. Retrieved 31 May 2016, from http://www.fifa.com/worldcup/news/y=2010/m=12/news=study-reveals-tourism-impact-south-africa-1347377.html

Szymanski, S. (2002). The Economic Impact of the World Cup. World Economics, 3(1), 169-177.

(31)