Why were voters against the 2022 Munich Winter Olympics in a referendum?
Coates, Dennis ; Wicker, Pamela
Introduction
The city of Munich (capital of the German federal state of Bavaria)
had submitted a bid for the 2018 Winter Olympics. Munich would have been
the first city ever to host Olympic Summer and Winter Games. However, it
lost against Pyeongchang, South Korea, which won in the first round of
voting. One issue with Munich's application was that several
farmers were against the bid because they had ecological concerns and
did not want to provide their land. Fifty-eight percent of voters from
Garmisch-Partenkirchen, which would host Alpine skiing events, voted in
favor of a petition in support of hosting the 2018 Winter Olympics only
two months before the International Olympic Committee (IOC) met to
decide the host for 2018; the timing hindered Munich's lobbying
efforts. At the same time only 51% voted against a review of the
contracts necessary to host the event (Mackay, 2013). This somewhat weak
public support was also observed by the IOC members and may have
influenced their decision. Public support is one of the requirements of
a bid and the bid committee of the IOC is required to take public
opinion into account when selecting an Olympic host city. For
Munich's 2018 bid no referendum had been held.
The supporters of the bid planned to submit another bid for the
2022 Winter Olympics--but this time with the prior approval of the
entire population of the host region. For this purpose, a referendum was
held on November 10, 2013. People living in four different regions were
allowed to vote: (1) city of Munich, (2) city of Garmisch-Partenkirchen
(which would host the alpine skiing, ski jumping, and Nordic combined
competitions), (3) district of Traunstein (including the community of
Koenigssee, which would host the bob, luge, and skeleton competitions),
and (4) district of Berchtesgadener Land (including the community of
Ruhpolding, which would host the biathlon and cross country skiing
competitions; Spiegel Online, 2013). In order for the referendum to
pass, the majority of people in all four regions would have to vote for
the bid. The outcome of the referendum was that all four regions voted
against the bid, although the share of yes votes differed among
communities: in Munich, 47.8% voted on favor of the bid, 48.4% in
Garmisch Partenkirchen, 40.3% in the district of Traunstein, and 45.9%
in the district of Berchtesgadener Land (Munich, 2013). The purpose of
this study is to identify the effect of various community
characteristics had on the percent of favorable votes in this
referendum. Secondary community-level data on voting outcomes and
community characteristics (i.e., proximity, socio-economic factors, and
political factors) are used for the examination. Specifically, this
study seeks to understand the determinants of voting on this referendum
to learn if there are lessons for the broader public debate on hosting
mega-events.
Referenda and Political Support
The present research question is a particularly interesting one in
the sports policy literature for at least two reasons. First, referenda
on hosting mega-events are rare because there is no culture of having
referenda in many democratic states (with the exception of Switzerland,
where referenda are common). Politicians are considered representatives
of the population and make such decisions on their behalf. At the same
time, the increasing spending that is associated with hosting
mega-events and the increasing concerns of the population that the money
could be spent elsewhere (e.g., health system, education) raises the
pressure on politicians who delegate the decision of whether the city
should submit a bid. Research has shown that many events were not found
to deliver the predicted positive economic welfare (e.g., Baade,
Baumann, & Matheson, 2010; Peeters, Matheson, & Szymanski,
2014). Another reason for the rareness of referenda could be that the
bidding committee tries to avoid them because they were often lost.
One of the first referenda on hosting the Olympics was held in the
context of Olympic Games in Denver. In 1970, Denver was awarded the 1976
Winter Olympics by the IOC. In 1972, a state-wide referendum was held
and a bond issue that should have helped finance the Games was rejected.
Consequently, Denver withdrew as an Olympic host (Hersh, 2015). In
recent years, referenda on hosting mega-events are becoming more common.
Indeed, there were at least four such referenda with regard to the 2022
Winter Olympics, in Krakow, Poland; Oslo, Norway; and St. Moritz/Davos,
Switzerland, as well as in Munich. Only the Norway bid passed the
referendum, while the Krakow bid and St. Moritz/Davos bid were not
approved by the public. Moreover, the bid of the 2010 Winter Olympics in
Vancouver, Canada, was preceded by a referendum that passed with 64% yes
votes. At the time of writing, Boston was bidding for the 2024 Olympic
Summer Games, but given the weak public support for this bid in recent
polls, the U.S. Olympic Committee (USOC) is concerned when it comes to a
referendum given the negative outcome of the Denver referendum (Hersh,
2015).
Second, the political parties in power--CSU (Conservative Party;
state government) and SPD (Social Democratic Party; city of Munich
government)--were strong supporters of the bid, yet the referendum
failed. For example, Horst Seehofer (CSU), the Prime Minister of the
state of Bavaria, supported the Olympic bid of Munich. He said that he
would support the Olympic bid and would also express this support in the
parliament (Focus, 2013). Christian Ude (SPD), the Mayor of Munich, also
argued in favor of submitting the bid. He encouraged everyone who would
be allowed to vote on that Sunday in Munich, Garmisch-Partenkirchen, in
the Berchtesgadener Land, or who lives in the district of Traunstein to
vote in favor of the Olympic idea and its chances. He stated that people
should vote for the most sustainable concept and against the giantism--
and against a monopoly of dictatorial regimes as a host of future
Olympic Games (German Olympic Sports Confederation [DOSB], 2013b).
Joachim Herrmann (CSU), the Bavarian Minister of the Interior, stated
further advantages of Munich hosting the 2022 Winter Olympics. He said
that the 2022 Olympic Games would lead to a huge impulse for the
development of mass, elite, and disability sport in Bavaria and that
people should fight together for Olympic Games (DOSB, 2013a). While
ruling party members supported the bid, members from opposition parties
(e.g., Green Party, Leftist Party) were against it. For example, Ludwig
Hartmann, leader of the parliamentary group of the Green Party in the
Bavarian state parliament, stated that the financial risks of hosting
Olympic Games and the destruction of the environment would be enormous
(Abendzeitung Munchen, 2013). Similarly, members of the Leftist Party
were against the bid and also supported the campaign NOlympia.
Olympic bids were also supported by leading politicians in other
countries. In March 2010, the Polish head of state, Lech Kaczynski,
stated he would like Poland to submit a bid for the 2022 Winter Olympic
Games. Krakow's Mayor, Jacek Majchrowski, would have liked to see
Krakow hosting the Games because he anticipated positive effects for the
region: "I regret that the referendum has put a definite end to
[...] the project that I considered to be very important for the
development of the whole region" (Washington Post, 2014, n. p.). In
Norway and Switzerland, the bid was mainly supported by leading
politicians, but more people seemed concerned about the bid because of
the financial burdens of hosting the Olympics. In Norway, one of the two
political parties in power, the Progress Party, had voted against a
state guarantee of 4 billion [euro], while the Conservative Party,
including the Norwegian Prime Minister Erna Solberg, supported
Oslo's bid (n-tv, 2014). Tarzisius Caviezel, Mayor of Davos,
Switzerland, supported the bid, but expressed his frustrations because
the critics would systematically frighten the population with
calculations of the anticipated debts accompanied with hosting the
Olympics (Tagesanzeiger, 2013).
Related Literature
Referenda in Sport
Previous research on referenda in sport has mainly looked at
referenda in the context of building sport stadia in North America
(e.g., Brown & Paul, 1999; Coates & Humphreys, 2006; Fort,
1997). One study examined the referendum on the location of a new
professional football (soccer) stadium (Allianz Arena) that was held in
Munich in 2001. Results showed that voters generally supported the
construction of the new arena, but tried to shift the stadium away from
their neighborhood (Ahlfeldt & Maennig, 2012). Yet, in the case of
the Allianz Arena only the infrastructure (e.g., roads, public
transport) was paid by the local government, not the facility itself.
Moreover, several studies analyzed people's willingness to support
bids for hosting Olympic Games (e.g., Atkinson, Mourato, Szymanski,
& Ozdemiroglu, 2008; Preuss & Werkmann, 2011; Walton, Longo,
& Dawson, 2008) using the contingent valuation method (CVM).
However, previous research has not yet looked at referenda for hosting
major sport events like the Olympic Games, possibly because referenda
are relatively new in this context.
Factors Determining the Decision of Voters
The individual decision to vote for or against a bid for a
mega-sport event can be conceptualized like a cost-benefit-analysis
(Kesenne, 2005) or a utility maximization problem (Coates &
Humphreys, 2006). Individuals may carefully think about the benefits and
costs that hosting the Olympic Games would have for themselves and
possibly also for the wider community or region. Their votes may be
driven by individual perceptions of utilities in various areas. For
example, an increase in public debt as a result of hosting the event may
decrease the individuals' utility (Feld & Kirchgassner, 2001)
and people may decide to vote against the bid. Altogether, it is assumed
that the voters' decisions in the present research are influenced
by a variety of factors that can be classified into at least three
areas: (1) proximity, (2) socio-economic factors (e.g., population
structure, importance of tourism, taxes, employment situation), and (3)
political factors. These factors are explained in the next paragraphs.
First, it is relevant whether people live in close proximity to the
event and are, thus, directly affected by the event. Applied to Olympic
bids, it is important whether the community the individual is living in
would host some of the events. On the one hand, people could demonstrate
a "not in my backyard" attitude that was observed in the
context of the Allianz Arena in Munich (Ahlfeldt & Maennig, 2012).
On the other hand, people may appreciate the excitement of living close
to the location of a megaevent and, thus, support the bid (Atkinson et
al., 2008). Coates and Humphreys (2006) analyzed referenda for two sport
facilities in the United States. They found evidence for proximity
benefits: voters in close proximity to the facilities tended to support
subsidies more than people living farther away from the facilities. For
the present study it is difficult to formulate an assumption: while it
looks like the majority of studies support a positive effect of
proximity on the support of the event (e.g., Atkinson et al., 2008;
Coates & Humphreys, 2006), a recent study on voting behavior in
Munich--the geographic area of interest--found negative proximity
effects (Ahlfeldt & Maennig, 2012).
Second, socio-economic factors including the population structure,
importance of tourism, taxes, and employment situation may drive the
decision to vote in favor or against the bid. Starting with the
population make-up, the gender and age distribution may be critical.
Research shows that males and younger people are more interested in
sports and more likely to participate in sport themselves (e.g.,
Downward & Rasciute, 2015). Walton et al. (2008) documented a
positive effect of male gender and a negative age effect when examining
the determinants of willingness-to-pay (WTP) for hosting the 2012
Olympic Games. These results indicate that males and younger people
provided more support for the Games given their significantly higher
WTP. Yet, no significant effects of age and gender were found in
Atkinson et al.'s (2008) study examining the same event. Similarly,
Preuss and Werkmann (2011) found no significant effects of age and
gender when examining the WTP for hosting the 2018 Winter Olympics in
Munich. Given the inconsistent findings in previous research, the
formulation of assumptions for the present study is
difficult--particularly because the previous Munich study (Preuss &
Werkmann, 2011) documented insignificant effects.
Moreover, the importance of tourism to the community may play a
role in voting behavior. Specifically, if the community is a tourism
destination, it would expect a greater demand for tourism services, like
hotel accommodations and meals in restaurants, and thus may be more
supportive of a bid to host a major tourism attraction. Indeed, some
research shows that the tourism sector benefits from hosting major sport
events (Daniels, Norman, & Henry, 2004). Yet, other studies indicate
that the tourism benefits are often overstated (e.g., Baade et al.,
2010; Peeters et al., 2014). For example, for the 2002 Winter Olympic
Games in Salt Lake City, Utah, it was documented that not all sectors of
the economy benefited equally from the Games. While the hospitality
industry prospered, retailers (e.g., merchandise and department stores)
suffered and their experienced losses were greater than the gains of
hotels and restaurants (Baade et al., 2010). During the 2010 Football
World Cup, South Africa experienced a large increase in net tourism
during the event; yet, the net increase in tourists visiting the country
was less than half compared with the predictions of the event organizers
(Peeters et al., 2010). An overstatement of economic benefits is a
common phenomenon in economic impact studies that are conducted ex ante.
Ex-post studies are more likely to lead to realistic results because
they can rely on observed expenditure, travels, tax income, spectators,
etc. (Gratton et al., 2000; Porter & Fletcher, 2008).
While residents may appreciate hosting the Olympics, they may be
deterred by other factors that may be accompanied by the event. Franck
(2005) has shown that the reasons why referenda are rejected do not
necessarily have to do with the actual decision due for approval, but
more with the voters' subjective anticipated consequences. In the
context of the Olympics, tourist venues may be concerned with using
regular tourism--an assumption supported by a study on the 2002 Winter
Olympics when Utah ski resorts actually lost sales during the event
period (Baade et al., 2010). In this case, Olympic tourists may have
crowded out regular tourism. Crowding out of tourists is a common
phenomenon and has been observed in other contexts, too (e.g., Chou,
Hsieh, & Tseng, 2014). Since many of the voting communities are
tourist destinations, crowding out of regular tourists may be a concern
of local voters.
Furthermore, tax competition can have an impact on voting outcomes
in referenda (Feld, 1997). While economic benefits of events or
facilities are conferred to a small set of actors, the costs are
distributed among the public (Brown & Paul, 1999). In most cases,
taxpayer money is used to build facilities and organize major sport
events (e.g., Brown & Paul, 1999; Coates & Humphreys, 2006).
Moreover, it can be interesting to know where taxpayer money comes from.
Specifically, the type and amount of taxes a community generates may
reflect the importance of various industry branches to the community.
For example, communities generating high revenues from sales and
business taxes tend to have many business companies, stores, or hotels.
Given the evident discrepancy in tourism benefits between the
hospitality industry and retailers (that are both covered by sales
taxes) in previous research (Baade et al., 2010); it is difficult to
make assumptions as to how the importance of different taxes within a
community affects voting outcomes.
Moreover, the employment situation in the community must be
considered from a socio-economic perspective. Employment effects of
major sport events have been extensively examined in the literature
(e.g., Coates & Humphreys, 2003; Hotchkiss, Moore, & Zobey,
2003; Jasina & Rotthoff, 2008). The majority of studies show that
employment effects of sport events can be considered relatively small at
best (e.g., Baade & Matheson, 2000; Coates & Humphreys, 2003)
and are only short-term in nature (Spilling, 1999). For example,
Feddersen and Maennig (2013) did not find any significant employment
effects when examining the 1996 Atlanta Olympic Games. In their
analysis, they contradict the results of Hotchkiss et al. (2003), who
reported a substantial increase in jobs in the regions associated with
the Games. Thus, a few studies such as the one above nurture the hope of
local residents suffering from unemployment to find a new job in the
context of the event. Independent of the scientific research in this
field (which many voters may not be aware of), residents may perceive
the event as an opportunity for creating new jobs. Therefore, it could
be assumed that people tend to vote in favor of the Olympic bid in
communities with high unemployment.
Third, the voting outcome may be affected by political factors.
Generally speaking, decisions that are put to a vote are oftentimes
complex in nature and in their entirety difficult to understand for the
local population (Franck, 2005). Similarly, the decision to support an
Olympic bid is complex in nature. Since individuals cannot gather all
the information themselves, they have to trust in part what some of the
people in charge (i.e., leading politicians) say. As outlined previously
in this paper, financial risks and ecological concerns associated with
hosting the Olympics are typically mentioned by opposition parties,
while ruling party members tend to advance the positive effects of
hosting the Olympics in terms of economic impact, image improvements,
and sport development. Given these contradictory statements, people may
trust the members of the party they usually give their vote. Thus, the
voting behavior in political elections (e.g., elections of the state
parliament) may represent an indicator for the outcome of a referendum.
This implies that in communities with high support of the ruling
political parties, the percentage of yes votes in the referendum may
also be higher compared to communities where opposition parties receive
a high share of the votes. As mentioned earlier, the ruling parties at
the time of the referendum were the Conservative Party (Bavaria state
government) and the Social Democratic Party (Munich city government). It
can therefore be expected that the higher the support for these two
ruling parties in the community, the higher the share of yes-votes in
the referendum.
Data and Empirical Approach
Data Collection
To examine the factors that affected the voting outcome in the
communities, secondary data on all 52 communities that participated in
the referendum were collected. These communities include the cities of
Munich and Garmisch-Partenkirchen, 35 communities in the district of
Traunstein, and another 15 communities in the district of
Berchtesgadener Land. Munich has more than 1.4 million inhabitants
across 25 neighborhoods (i.e., suburbs) for which voting results are
available and is by far the most populous community in the region. We
consider two distinct samples, one that treats the 52 communities as
equally influential observations and one of 76 communities in which the
25 neighborhoods of Munich are treated as separate observations. This
last approach is feasible because the neighborhoods in Munich have
several thousand inhabitants and thus are far larger than most of the
more rural communities in the data.
The percentage of yes votes by community was made available from
the official website of the city of Munich (Munich, 2013). Moreover,
data on population numbers by age and gender, tourism, unemployment, and
tax income were retrieved for every community from the official website
of the German Federal Statistical Office (Federal Statistical Office,
2014). The respective population, tourism, and unemployment data for the
25 suburbs of Munich were collected from the website of the Statistical
Office of Munich (Statistical Office Munich, 2014a; 2014b; 2014c); tax
data are not available on the suburb level and so are other data on
household characteristics that are only available at a district or state
level. Thus, the sample with 76 communities in which the 25
neighborhoods of Munich are treated as separate observations limits the
number of explanatory variables because of data availability issues.
Empirical Analysis
Regression models (OLS models with robust standard errors) examine
the effect of various community characteristics on the referendum voting
results. The dependent variable is the percent of favorable votes in
each community. Several independent variables are included in the
analysis. Since not all communities involved in the referendum would
also be hosts, one variable captures whether the community is a
potential host (1=yes). Potential tourism impact is proxied with the
number of hotel beds in the community per capita. The labor market
situation is measured with the number of unemployed people per capita.
Two months before the referendum state election were held in Bavaria.
The shares for the top five political parties are also included in the
models to measure the political orientation of inhabitants. These are
the Conservative Party (CSU), the Social Democratic Party (SDP), the
Green Party (Bundnis 90/Die Granen), the Liberal Party (FDP), and the
Leftist Party (Die Linke). The models also control for population
characteristics including the share of males and the share of
inhabitants of working age (18 to 64 years). Three sets of models are
estimated. In the first set, all independent variables from Table 1 are
included. In the second set, the models contain the best specification
variables that were selected based on the corrected Akaike Information
Criterion (AICC). Both sets of models are estimated for the larger
sample (n=76; including the 25 Munich neighborhoods) and for the
restricted sample (n=52; Munich represents one observation). In the
third set, the best specification for the three regions (Munich,
Traunstein, Berchtesgadener Land) using the AICC will be presented. (1)
An -level of 0.1 is used for all statistical tests.
Results and Discussion
Descriptive statistics are reported in Table 1 for both samples.
Because Munich is far more populous than any of the other voting
districts, all explanatory variables are measured as percentages,
proportions, or per capita. Although there is little difference between
the two samples, some variables are worth noting. One of these is the
potential host. Only one of the 25 neighborhoods of Munich is identified
as a potential host of some Winter Olympic event; so in the 76
observations sample the proportion of potential hosts is about a third
smaller than in the sample when Munich is a single observation. Also,
the shares of the political parties differ between the two samples. In
the sample in which Munich is a single observation the Conservative
share is higher, while the Social Democrat share is lower than in the
sample in which the 25 Munich neighborhoods are treated as separate
observations. This is because the inhabitants of Munich favor the Social
Democratic Party (and also have the mayor from this party), while people
in the rest of Bavaria tend to support the Conservative Party. Thus, the
Social Democrat share is higher when Munich is represented with 25 cases
in the sample. The tax variables are not available for the individual
neighborhoods of Munich. Consequently, our analysis can only use tax
variables when the sample treats Munich as a single observation rather
than as 25 separate voting districts. Tax variables are measured as
thousands of Euros per capita. For example, the mean income tax revenue
per capita of 0.367 indicates an average of 367 [euro] per person.
Results of regression models are reported in Tables 2 through 5.
Table 2 shows two models, one using 76 observations, in which the 25
Munich neighborhoods are separate observations, and another with 52
observations with Munich as just one observation. The results between
the two samples are quite similar. Many variables have roughly equal
coefficient estimates in both samples, but often the estimate is only
significant in the larger sample, if at all. In both samples, the
stronger is the Leftist Party in a community, measured as the share of
the vote from the previous federal state election, the lower is the
likelihood of a vote in favor of the Munich Winter Olympics bid.
Likewise, in both samples, the larger is the Green Party share of the
vote, the lower is the likelihood of a vote for the bid. Thus, the
effects of the opposition parties' share are significant and
negative, while the effects of the ruling parties' share are
inconsistent and mostly insignificant. The positive and significant
effect of the potential host variable shows that districts identified as
potential hosts of some portion of the event have a higher likelihood of
a favorable vote. The positive effect of potential host is contrary to
previous research examining voting behavior at a referendum in Munich
where voters demonstrated a "not in my backyard" attitude
(Ahlfeldt & Maennig, 2012).
A higher level of unemployment, as measured by the number of
unemployed people per capita in the community, leads to a statistically
significant increase in the likelihood of a favorable vote in the larger
sample, but not in the smaller. Unemployed people may assume that
employment opportunities are associated with the Olympics. The effect is
only significant in the larger sample because unemployment rates are
typically higher in large cities than in rural areas and the effect of
Munich is stronger when its 25 neighborhoods are considered separately.
In the larger sample, the higher the number of hotel beds per capita in
a district, the smaller the likelihood of a yes vote. The initial
hypothesis was that hotel beds per capita is a measure of the strength
of tourism in the economy and would indicate support for the bid. Our
results are at odds with that assumption. It is likely that the hotels
in the regions are already well occupied in the winter and people may be
afraid of losing their regular guests during the period of the Olympic
Games. The concern of crowding out of regular tourists is nurtured by
previous research (Baade et al., 2010; Chou et al., 2014).
A difficulty in voting studies of this sort is that the variables
are all imperfect proxies for the determinants of individual voting
behavior. Moreover, variables may be highly correlated with one another,
leading to weak results from individual hypothesis tests. Consequently,
joint significance tests are conducted on variables in groups. Table 2
reports the F-statistic and p-value for the null hypotheses that (a) all
political party variables carry coefficients of zero, (b) all
socio-economic variables' coefficients are zero (excluding the tax
variables), (c) the three regional dummy variables have zero
coefficients, and (d) all tax variables have zero coefficients in the
restricted sample model. The alternative hypothesis in each case is that
at least one of the coefficients in the group is non-zero. For this
test, the tax variables are considered separately because they are not
available for the larger sample including the 25 Munich neighborhoods.
The results in Table 2 indicate that one can reject the null hypothesis
for the political party and regional dummy variables in both models, but
cannot reject the null hypothesis with respect to tax variables and the
remaining socio-economic variables.
To aid interpretation, Table 3 presents elasticity estimates for
the coefficients from the regression equations of Table 2. For example,
a 10% increase in unemployment share raises the yes vote percentage by
1.28%. At the mean of the percent of favorable votes (M=44.1), 1.28% is
about 0.56 percentage points of vote share; the share of favorable votes
is not especially responsive to changes in the unemployment share. A 10%
increase in the share of Leftist votes would decrease the percentage of
yes votes by 1.79% (larger sample) and 1.98% (smaller sample),
respectively. The corresponding percentage points of vote share at the
mean of the percent of favorable votes are 0.79 (larger sample) and 0.87
(smaller sample), respectively. For the share of Green party votes, the
respective percentage points of vote share at the mean of the percent of
favorable votes are 0.74 (larger sample) and 1.08 percentage points
(smaller sample). All of the elasticities are relatively small,
indicating that favorable votes are only weakly responsive to the
explanatory variables in this first set of models.
Table 4 reports models selected using the AICC. Each model was
forced to include the potential host variable, and to select the best
model from all the other possible combinations of the explanatory
variables. The included regional dummies also come from the best fitting
model according to the smallest corrected AIC value. Which variables are
individually significant is similar across Tables 2 and 4. Coefficient
estimates are also similar. It is not surprising then that the
elasticity estimates are also quite similar.
Table 5 reports results of estimating the model on observations
only from a given region and the corresponding elasticities. In effect,
this allows every coefficient in the model to differ from region to
region. Unfortunately, the city of Garmisch Partenkirchen is not
included as it is a single observation. Moreover, none of the regions
has many observations, with Traunstein the most with 35. As before,
there are no tax variables for the separate neighborhoods of Munich so
the search for the best Munich model based on the corrected AIC does not
include taxes. The models for Traunstein and Berchtesgadener Land
include all the variables.
The Munich regression is interesting for several reasons. First,
potential host is not significant. This is likely explained by the fact
that there is only one neighborhood within Munich that might have hosted
any Winter Olympic events. The second interesting aspect of the Munich
regression is that the percent of the working population, the share of
males, and the Conservative share have a significant positive effect,
but have been insignificant before. The support of younger people and
males is in line with previous research (Walton et al., 2008). The
Social Democrat share is also significant. It had been negative and
significant at the 10% level when Munich was entered as a single
observation in Table 2, but now it is significant at the 5% level and
positive. The elasticities of yes percentage with respect to these
variables are among the largest, with the male share elasticity close to
one.
Regressions for Traunstein and Berchtesgadener Land are also
interesting. In each case, being a potential host raises the likelihood
of a favorable vote. Also, the higher the proportion of Leftist Party
votes in the previous parliamentary election, the smaller the share of
favorable votes. The elasticity for Leftist share is similarly small for
Berchtesgadener Land as for Munich, both of which are about half the
elasticity of Leftist share in Traunstein, which is also quite small.
Oddly, a greater proportion of males in Berchtesgadener Land induces a
smaller proportion of voters favoring the bid, while a larger proportion
of males elsewhere, in Munich or Traunstein, raises the proportion
favoring it. The elasticity for Traunstein is 1.91, while for
Bechtesgadener Land it is -1.65. These elasticities are quite large,
suggesting a 10% increase in the male population raises the yes vote
percentage by 19% in Traunstein and lowers it by 16% in Berchtesgadener
Land. The share of the vote captured by the Green Party in the most
recent parliamentary election is negatively related to support for the
bid in Traunstein. It is interesting that Green share was significant in
Table 2 in both models, but is only significant in the Traunstein model
in Table 5. Also, only in Traunstein, higher sales tax revenues per
capita raise the percentage of votes favoring the bid, but higher
business taxes net of reallocations per capita lower the percentage of
favorable votes. Yes-vote share is inelastic with respect to either type
of tax.
Conclusion
The purpose of this study was to evaluate the determinants of the
percent of yes votes in a referendum on whether the city of Munich and
the state of Bavaria should bid to host the 2022 Winter Olympic Games.
Referenda on hosting mega-events are a recent phenomenon, and ours is
the first empirical analysis of such a vote of which we are aware. A
previous bid attempt had failed in part because the IOC observed
opposition based on ecological concerns among the population.
Consequently, the supporters of a bid sought a referendum to document
the local backing for a bid. The political parties in power in the state
of Bavaria and the city of Munich, the CSU (Conservative) and the SPD
(Social Democrat), supported the bid. Despite this, the referendum
failed, not receiving a majority of the votes in any of the four voting
districts, even Garmisch-Partenkirchen where a petition had passed in
support of a bid only three years earlier. Thus, the size of specific
opposition parties is important to voting outcomes. Our results show
that likely host communities tended to be more supportive of the bid. In
communities with greater Green Party and Leftist Party support, both of
which opposed the bid, voters were less likely to cast votes for the
bid. There is also evidence that communities with greater unemployment
were more supportive of the bid, while communities with a high number of
hotel beds per capita were less supportive.
There are lessons for other potential bidders in the results. The
findings indicate what factors are considered critical to the local
population when it comes to such voting decisions. Local politicians and
bidding committees could use this information to better understand the
local population and to improve their support for hosting Olympic Games
and other major sporting events. First, the positive effect of potential
host on the percent of yes votes suggests that winning support for the
bid to host the Winter Olympics is more likely the greater the dispersal
among the communities of possibly hosting an event. Second, the strength
of the influence of the share of the population at working age and the
male population in the analysis by region suggests that Olympic
promoters should target people over 65 years and females to build
support for the events. Third, the negative effect of hotel beds
indicates that the local population is concerned with a potential
crowding out of regular tourists during the Games--a concern that should
be addressed by the promoters of the bid. Finally, the consistent
strength of the unemployment variable indicates that bid supporters are
right to emphasize the employment possibilities linked with the bid.
The present research has some limitations that may guide future
research. It is based on a relatively small sample size, although
additional observations for neighborhoods in Munich were already added.
Yet, the sample size is naturally restricted to those communities
involved in the referendum. Moreover, the study is limited to the
secondary data that are available at the community level or neighborhood
level in Munich. Future research may examine voting behavior in
referenda beyond the community level, for example, by surveying citizens
in the respective regions. In this context it would be interesting to
examine what role potential financial burdens of bidding for Olympic
Games and hosting Olympic Games that are imposed by the IOC play in the
decision-making process of citizens.
Dennis Coates is a professor of economics at the University of
Maryland, Baltimore County, and an international researcher at HSE-Perm,
Russian Federation. His research interests include sport economics and
political economy.
Pamela Wicker is a senior lecturer at the German Sport University
Cologne. Her research interests include economics of sport consumption,
spectator sport economics, and non-profit economics.
References
Abendzeitung Munchen. (2013). Grune gegen
Olympiabewerbung--"Sind wir denn Deppen?". Retrieved from
http://www.abendzeitung-muenchen.de/inhalt.winterspiele-2022-inmuenchen-gruene-gegen-olympia-
bewerbung-sind-wir-denn-deppen.d17e56c9-b116-4c03-b785-118204f4f40f.html
Ahlfeldt, G., & Maennig, W. (2012). Voting on a NIMBY facility:
Proximity cost of an "iconic" stadium. Urban Affairs Review,
48, 205-237.
Atkinson, G., Mourato, S., Szymanski, S., & Ozdemiroglu, E.
(2008). Are we willing to pay enough to 'back the bid'?
Valuing the intangible impacts of London's bid to host the 2012
Summer Olympic Games. Urban Studies, 45, 419-444.
Baade, R. A., Baumann, R. W., & Matheson, V. A. (2010).
Slippery slope? Assessing the economic impact of the 2002 Winter Olympic
Games in Salt Lake City, Utah. Region et Developpement, 31, 81-91.
Baade, R. A., & Matheson, V. A. (2000). An assessment of the
economic impact of the American football championship, the Superbowl, on
host communities. Reflets et Perspectives de la Vie Economique, 39(2/3),
35-46.
Brown, C., & Paul, D. M. (1999). Local organized interests and
the 1996 Cincinnati sports stadia tax referendum. Journal of Sport &
Social Issues, 23, 218-237.
Coates, D., & Humphreys, B. R. (2003). The effect of
professional sports on earnings and employment in the services and
retail sectors in US cities. Regional Science and Urban Economics, 33,
175-198.
Coates, D., & Humphreys, B. R. (2006). Proximity benefits and
voting on stadium and arena subsidies. Journal of Urban Economics, 59,
285-299.
Chou, C., Hsieh, S. F., & Tseng, H. P. (2014). The crowding-out
effects of Chinese tourists on inbound tourism in Taiwan. Tourism
Economics, 20, 1235-1251.
Daniels, M. J., Norman, W. C., & Henry, M. S. (2004).
Estimating income effects of a sport tourism event. Annals of Tourism
Research, 31, 180-199.
DOSB. (2013a). Bayerns Innenminister fur Olympia 2022 in Munchen.
Retrieved from http://www.dosb.de/de/olympia/olympische-news/detail/news/bayerischer_innenminister_fuer_olympia_2022_in_muenchen/
DOSB. (2013b). Munchen 2022: Christian Ude widerspricht den
Kritikern. Retrieved from
http://www.dosb.de/de/olympia/olympische-news/detail/news/-83aa84424d/
Downward, P., & Rasciute, S. (2015). Exploring the covariates
of sport participation for health: An analysis of males and females in
England. Journal of Sports Sciences, 33, 67-76.
Feddersen, A., & Maennig, W. (2013). Employment effects of the
Olympic Games in Atlanta 1996 reconsidered. International Journal of
Sport Finance, 8, 95-111.
Federal Statistical Office. (2014). Regionalstatistik. Retrieved
from https://www.regionalstatistik.de/genesis/online
Feld, L. P. (1997). Exit, voice, and income taxes. European Journal
of Political Economy, 13, 455-478.
Feld, L. P., & Kirchgassner, G. (2001). Does direct democracy
reduce public debt? Evidence from Swiss municipalities. Public Choice,
109, 347-370.
Focus. (2013). Kabinett unterstutzt Olympiabewerbung. Retrieved
from http://www.focus.de/regional/muenchen/kabinett-kabinett-unterstuetzt-olympiabewerb ung_aid_1135852.html
Fort, R. (1997). Direct democracy and the stadium mess. In R. G.
Noll & A. Zimbalist (Eds.), Sports, jobs and taxes: The economic
impact of sports teams and stadiums (pp. 146-177). Washington, DC: The
Brookings Institution Press.
Franck, R. (2005). Why did a majority of French voters reject the
European Constitution? European Journal of Political Economy, 21,
1071-1076.
Gratton, C., Dobson, N., & Shibli, S. (2000). The economic
importance of major sports events: A case-study of six events. Managing
Leisure, 5, 17-28.
Hersh, P. (2015). Could vote kill Boston Olympic bid? Chicago
Tribune. Retrieved from
http://www.chicagotribune.com/sports/international/chi-could-vote-kill-boston-olympicbid-20150120-story.html
Hotchkiss, J. L., Moore, R. E., & Zobey, S. M. (2003). Impact
of 1996 Summer Olympic Games on employment and wages in Georgia.
Southern Economic Journal, 69, 691-704.
Jasina, J., & Rotthoff, K. W. (2008). The impact of a
professional sports franchise on county employment and wages.
International Journal of Sport Finance, 3, 210-227.
Kesenne, S. (2005). Do we need an economic impact study or a
cost-benefit analysis of a sports event? European Sport Management
Quarterly, 5, 133-142.
Mackay, D. (2013). Munich sets date for referendum on whether to
bid for 2022 Winter Olympics and Paralympics. Retrieved from
http://www.insidethegames.biz/olympics/winterolympics/2022/1014547-munich-sets- date-for-referendum-on-whether-to-bid-for-2022-
winter-olympics-and-paralympics
Munich. (2013). Burgerentscheid "Bewerbung um Olympische und
Paralympische Winterspiele 2022" am 10.11.2013. Retrieved from
http://www.muenchen.de/rathaus/Stadtpolitik/
Wahlergebnisse/Buerger_Volksentscheide/Winterspiele_2022.html
n-tv. (2014). Winterspiele 2022 in Oslo? Parlament streitet um
Olympia-Bewerbung. Retrieved from
http://www.n-tv.de/sport/olympia/Parlament-streitet-um-Olympia-Bewerbung-article12771196.html
Peeters, T., Matheson, V. A., & Szymanski, S. (2014). Tourism
and the 2010 World Cup: Lessons for developing countries. Journal of
African Economies, 23, 290-320.
Porter, P. K., & Fletcher, D. (2008). The economic impact of
the Olympic Games: Ex ante predictions and ex post reality. Journal of
Sport Management, 22, 470-486.
Spiegel Online. (2013). Olympia-Bewerbung 2022: Biathlon und
Langlauf sollen in Ruhpolding stattfinden. Retrieved from
http://www.spiegel.de/sport/wintersport/olympia-bewerbung2022-biathlon-und-langlauf- in-ruhpolding-a-921866.html
Spilling, O. R. (1999). Long-term impacts of mega-events: The case
of Lillehammer 1994. In C. Jeanrenaud (Ed.), The economic impact of
sport events (pp. 135-166). Neuchatel: Editions CIES.
Statistical Office Munich. (2014a). Arbeitsmarkt. Retrieved from
http://www.muenchen.de/ rathaus/Stadtinfos/Statistik/Arbeitsmarkt.html
Statistical Office Munich. (2014b). Bevolkerungsbestand. Retrieved
from http://www.muenchen.
de/rathaus/Stadtinfos/Statistik/Bev-lkerung/Bev-lkerungsbestand.html
Statistical Office Munich. (2014c). Wirtschaft. Retrieved from
http://www.muenchen.de/ rathaus/Stadtinfos/Statistik/Wirtschaft.html
Tagesanzeiger. (2013). Den Bundnern wird systematisch Angst
gemacht. Retrieved from
http://www.tagesanzeiger.ch/schweiz/standard/Den-Buendnern-wird-systematisch- Angstgemacht/story/27754249?dossier_id=1620
Walton, H., Longo, A., & Dawson, P. (2008). A contingent
valuation of the 2012 London Olympic Games. Journal of Sports Economics,
9, 304-317.
Washington Post. (2014). Krakow withdraws 2022 Olympic bid after
residents vote 'no'. Retrieved from
http://www.washingtonpost.com/blogs/early-lead/wp/2014/05/27/krakow-withdraws2022-olympic-bid-after- residents-vote-no/
Endnote
(1) At the suggestion of a referee, separate models were estimated
that include only one group of variables, political, socio-economic, or
regional, as well as the potential host dummy. These results are
available upon request
Dennis Coates [1,2] and Pamela Wicker [3]
[1] University of Maryland, Baltimore County
[2] National Research University, Higher School of Economics
[3] German Sport University Cologne
Table 1. Descriptive Statistics
Obs. Mean Std. Minimum Maximum
Dev.
Variables 25 Munich neighborhoods separate
observations
Percent of favorable votes 76 44.106 6.552 24.740 57.370
Potential host (1 = yes) 76 0.053 0.225 0.000 1.000
Conservative share 76 44.735 9.639 23.780 62.069
Social Democrat share 76 19.127 9.841 4.987 38.230
Green share 76 14.201 6.310 4.894 38.004
Liberal share 76 4.706 2.289 0.911 10.110
Leftist share 76 2.268 1.020 0.342 4.859
Hotel beds per capita 76 0.126 0.227 0.000 1.320
Percent age 18 to 64 76 63.689 4.667 55.888 79.400
Proportion male 76 0.492 0.013 0.457 0.517
Unemployed per capita 76 0.018 0.007 0.007 0.037
Munich one observation
Percent of favorable votes 52 42.364 6.889 24.740 57.370
Potential host (1 = yes) 52 0.077 0.269 0.000 1.000
Conservative share 52 48.612 8.186 36.053 62.069
Social Democrat share 52 13.061 4.383 4.987 32.127
Green share 52 15.065 6.940 4.894 38.004
Liberal share 52 4.353 2.425 0.911 9.310
Leftist share 52 2.212 1.167 0.342 4.859
Hotel beds per capita 52 0.163 0.263 0.000 1.320
Percent age 18 to 64 52 61.500 2.158 55.888 67.534
Proportion male 52 0.492 0.013 0.457 0.517
Unemployed per capita 52 0.014 0.005 0.007 0.027
Income tax per capita 52 0.367 0.057 0.247 0.597
Sales tax per capita 52 0.029 0.020 0.005 0.102
Business tax minus 52 0.328 0.338 0.001 0.037
reallocation per capita
Farm property tax per 52 0.012 0.008 0.000 0.037
capita
Real property tax per 52 0.111 0.051 0.051 0.3111
capita
Table 2. Determinants of the Percent of Favorable Votes
Variables (1) (2)
25 Munich 1 Munich Obs.
Neighborhoods
Potential host 7.134 * (0.073) 11.09 *** (0.001)
Conservative share 0.00417 (0.981) -0.159 (0.339)
Social Democrat share -0.292 (0.263) -0.478 * (0.099)
Green share -0.491 ** (0.010) -0.646 *** (0.005)
Liberal share 0.328 (0.358) 0.278 (0.612)
Leftist share -3.305 *** (0.003) -3.541 *** (0.003)
Hotel beds -6.748 ** (0.024) -5.357 (0.288)
Percent age 18 to 64 0.214 (0.275) -0.290 (0.653)
Male share -8.101 (0.853) 0.297 (0.998)
Unemployed per capita 322.7 ** (0.022) 377.6 (0.255)
City of Garmisch -14.71 *** (0.008) -3.470 (0.700)
Partenkirche
Traunstein -3.375 (0.385) 11.58 (0.191)
Berchtesgadener Land -5.860 (0.107) 10.78 (0.270)
Income tax per capita 24.22 (0.334)
Sales tax per capita 155.9 (0.130)
Business tax minus -4.571 (0.178)
reallocation per capita
Farm property tax per 185.8 (0.116)
capita
Real property tax per -33.65 (0.400)
capita
Constant 50.46 * (0.073) 63.80 (0.367)
Observations 76 52
R-squared 0.695 0.741
(a) F_political 14.001 4.871
Prob>F 0.000 0.002
(b) F_socio-economic 1.854 0.944
Prob>F 0.130 0.451
(c) F_region 4.416 3.266
Prob>F 0.007 0.033
(d) F_tax 1.232
Prob>F 0.316
Note: *** p < 0.01; ** p < 0.05; * p < 0.1; robust
p-values in parentheses.
Table 3. Elasticities
(1) (2)
Variables 25 Munich 1 Munich Obs.
Neighborhoods
Potential host 0.007 * 0.016 ***
(0.050) (0.001)
Conservative share 0.004 -0.183
(0.981) (0.340)
Social Democrat share -0.125 -0.149 *
(0.263) (0.098)
Green share -0.168 ** -0.245 ***
(0.012) (0.005)
Liberal share 0.036 0.030
(0.359) (0.612)
Leftist share -0.179 *** -0.198 ***
(0.004) (0.003)
Hotel beds -0.020 ** -0.021
(0.029) (0.291)
Percent age 18 to 64 0.313 -0.429
(0.275) (0.653)
Male share -0.092 0.004
(0.853) (0.998)
Unemployed per capita 0.128 ** 0.126
(0.022) (0.257)
City of Garmisch -0.004 *** -0.001
Partenkirchen (0.008) (0.700)
Traunstein -0.039 0.196
(0.384) (0.194)
Berchtesgadener Land -0.026 0.069
(0.109) (0.270)
Income tax per capita 0.214
(0.336)
Sales tax per capita 0.105
(0.124)
Business tax minus -0.036
reallocation per capita (0.179)
Farm property tax per 0.054
capita (0.115)
Real property tax per -0.089
capita (0.400)
Observations 76 52
Note: *** p < 0.01; ** p < 0.05; * p < 0.1;
p-values in parentheses
Table 4. Best Fit and Elasticities--AICC
(1) (1)
Variables 25 Munich 25 Munich
Neighborhoods Neighborhoods
(Elasticities)
Potential host 6.918 * 0.007 **
(0.063) (0.044)
Leftist share -3.355 *** -0.181 ***
(0.000) (0.000)
Green share -0.409 *** -0.139 ***
(0.001) (0.001)
Munich 1.286 0.009
(0.471) (0.470)
Hotel beds -4.484 * -0.013 *
(0.077) (0.087)
Unemployed per capita 196.907 * 0.078 *
(0.092) (0.090)
Berchtesgadener Land -3.094 ** -0.013 **
(0.034) (0.039)
City of Garmisch -10.438 *** -0.003 ***
Partenkirchen (0.003) (0.003)
Traunstein
Male share
Constant 54.573 ***
(0.000)
Observations 76 76
R-squared 0.680
(2) (2)
Variables 1 Munich Obs. 1 Munich Obs.
(Elasticities)
Potential host 5.657 * 0.008 **
(0.050) (0.040)
Leftist share -2.899 *** -0.161 ***
(0.001) (0.001)
Green share -0.379 *** -0.143 ***
(0.007) (0.009)
Munich
Hotel beds
Unemployed per capita 329.600 0.110
(0.199) (0.197)
Berchtesgadener Land
City of Garmisch
Partenkirchen
Traunstein 3.531 * 0.059 *
(0.057) (0.053)
Male share 76.483 0.901
(0.230) (0.232)
Constant 9.414
(0.781)
Observations 52 52
R-squared 0.596
Note: *** p < 0.01; ** p < 0.05; * p < 0.1; (robust)
p-values in parentheses.
Table 5. Regions Best Fit and Elasticities--AICC
(1) (1)
Variables Munich Munich
(Elasticities)
Potential host -0.026 -0.001
(0.182) (0.183)
Percent age 18 to 64 0.021 *** 0.760 ***
(0.000) (0.000)
Male share 3.641 *** 0.938 ***
(0.003) (0.003)
Conservative share 0.031 *** 0.574 ***
(0.000) (0.000)
Hotel beds -0.425 ** -0.011 **
(0.029) (0.029)
Social Democrat share 0.013 ** 0.222 **
(0.018) (0.018)
Leftist share -0.086 ** -0.109 **
(0.016) (0.017)
Sales tax per capita
Business tax minus
rellocation per capita
Green share
Constant -4.642 ***
(0.000)
Observations 25 25
R-squared 0.953
(2) (2)
Variables raunstein Traunstein
(Elasticities)
Potential host 0.414 *** 0.005 ***
(0.000) (0.000)
Percent age 18 to 64
Male share 6.581 ** 1.911 **
(0.012) (0.013)
Conservative share
Hotel beds -0.366 *** -0.030 ***
(0.000) (0.001)
Social Democrat share
Leftist share -0.125 *** -0.206 ***
(0.000) (0.000)
Sales tax per capita 8.848 *** 0.140 ***
(0.001) (0.001)
Business tax minus -0.365 *** -0.075 ***
rellocation per capita (0.001) (0.001)
Green share -0.020 *** -0.219 ***
(0.003) (0.004)
Constant -3.012 **
(0.026)
Observations 35 35
R-squared 0.707
(3) (3)
Variables Berchtes- Berchtes-
gadener Land gadener Land
(Elasticities)
Potential host 0.460 *** 0.013 ***
(0.000) (0.000)
Percent age 18 to 64
Male share -6.116 *** -1.649 ***
(0.003) (0.003)
Conservative share
Hotel beds
Social Democrat share
Leftist share -0.135 *** -0.093 ***
(0.004) (0.004)
Sales tax per capita
Business tax minus
rellocation per capita
Green share
Constant 2.975 ***
(0.003)
Observations 15 15
R-squared 0.836
Note: *** p < 0.01; ** p < 0.05; * p < 0.1; (robust)
p-values in parentheses.
