Influence Chart & Establishing Objective

Economists complain that hosting the Olympics is financially irresponsible and that the costs vastly outweigh any tangible benefits.  If this is such an irrefutable claim, then why do public officials continue to push bids for their cities and nations?  I am interested in evaluating any advantages of hosting that have been previously undiscovered (or undervalued) by economists.  From a public policy perspective, I hope that my research will eventually prove a useful tool in determining whether or not a host nation will submit a bid.  As a Norwegian studies minor, I am especially interested in the 1994 Winter Olympics in Lillehammer as the Winter Olympics should (theoretically) be more successful in a country with both mountains and a climate that produces natural snow.  But Norway pulled out their most recent bid for the 2022 games due to lack of public support.  Perhaps if my research can pinpoint specific benefits of hosting the Olympics that previous research has missed, government officials and citizens will be more open to hosting the Winter Olympics and the negative trend in number of countries bidding will reverse, resulting in more bids for the IOC to choose from, better Olympic Games and increased benefits for host nations.  And if my research fails to see any real correlation between hosting the games and certain assumed benefits, economists will be further convinced that hosting is a financially irresponsible decision.

To begin my analysis, I created an influence chart to recognize the variables, parameters and their respective relationships with the ultimate and immediate outcomes in hosting the Winter Olympics.  The ultimate outcome is profit from a business perspective because without economic profit, the International Olympic Committee (IOC) would have a difficult time getting countries to bid on hosting the games.  From a public policy point of view, this study aims to incorporate myriad qualitative benefits (like increased happiness) from hosting the games that are often left out of simple economic analyses. The variables contributing to profit are total revenue and total cost.  Contributing to these two variables are a variety of fixed inputs and additional variables.  In the interest of readability, I included a simplified version of my influence chart below:

                     

The main components of total revenue are attendance and Olympic grants (any money that the IOC and private donors contribute to the host country).  The number of viewers and the games' attendance drive total revenue, but it is necessary to consider what portion of the attending individuals are local residents, as this group will have a lower effect on revenue in lodging and accommodations.  The attendance of the games will also theoretically affect media revenue, tourist revenue, ticket sales and revenue from international and domestic sponsorships.  These variables will be examined in greater detail later in my analysis.

To calculate the total cost of hosting the games, variable costs must be differentiated from fixed costs. For example, a fixed cost would be the initial payment to the IOC to be considered as a host country in the bidding process.  Variable costs from the bidding process would be the expense of consulting firms that the Norwegian government hired to create a presentation, brochures and propaganda material for IOC members and any costs associated with the wining and dining of the IOC.  Important costs to consider are construction and modernization costs, operation costs (including increased security, opening & closing ceremony) and opportunity costs which are often omitted from assessments.

Short-Term Forecasting Model: Television Broadcast Revenue

Television revenue is the biggest contributors to the host city's total revenue and has enjoyed steady growth over the lifespan of the Winter Olympics.  Host countries are entitled to approximately 60% of the television rights fees according to Denis Oswald of the IOC Executive Committee.  Selling the rights to the broadcast produces enormous profit and, as technology has developed, broadcasting has been able to reach increasingly large television audiences.  In the IOC's Olympic Marketing File from 2014, the broadcast revenue is given for each of the Winter Olympic games since 1960 in Squaw Valley.

The revenue is given in millions of US dollars and represents the total worldwide network commitment of rights fees.  These fees include both cash and technical service components. Lillehammer, in particular, saw $352.9 million in revenue.  It is important to note, however, that the service components are typically defrayed by the host nation (paid by the country not the television network) and that the broadcast revenue is not always paid in lump sum and can be divided into payments designated in long contracts with the television network.

Using the data from before the Sochi games, I created a short-term forecasting model using Holt's method to identify any trends in the data and forecast Sochi's eventual revenue.   Below is a graph comparing the observed and forecasted data points in each city since 1960.  The broadcast revenue enjoyed a spike for coverage of the Sarajevo Games in 1984, kickstarting an exponential growth trend that continues today.  This growth is probably due to increased numbers of viewers because of rising broadcast technology.  My model predicts $1.257 billion in broadcast revenue for Sochi and judging from the graph, that number will probably be a slightly lower than the observed value for 2014.

*The actual broadcast revenue figure for Sochi ended up being $1.26 billion which was slightly higher than my model's estimate but extremely close.

Non-Linear Optimization in Locating Sport & Non-Sport Venues

An important component of total cost in hosting the Winter Olympics is creating new infrastructure.  Both sport and non-sport venues built to host the games, must be well-planned geographically and built with forethought concerning their sustainability and possible usefulness in the future.  One of the ways that the host city Olympic planning committee can minimize costs is placing non-sport venues in optimal locations in relation to sport venues.  By using a non-linear optimization model, I calculated the optimal location for the most important non-sport venue, the Lillehammer Olympic Village.  By using the GPS coordinates from each sport venue, I ran an optimization model that produced the ideal Olympic Village location of 61.1254 latitude and 10.4803 longitude.  Entering the coordinates into Google Maps shows that the location is merely farmland.  Upon closer inspection, the Håkon Hall is extremely close to the ideal location.  

After researching the actual location and construction of the Lillehammer Olympic Village, my model's optimal coordinates made more sense.  The actual Olympic Village was established right behind the Håkon Hall and was a portable design intended to minimize cost and be disassembled at the games' conclusion.  The actual location was exactly where my model placed it but the land has since been converted to farmland.  

My non-linear optimization model reaffirmed that the location of the Lillehammer Olympic Village was optimal in relation to the locations of the sport venues.  More importantly, the analysis provided additional information about current uses of venues which I will tour in person this summer as an extension of my research.    

Logistic Regression: Actual Realization of Olympic Budgets

A major problem for Winter Olympic hosts is exceeding the original budget, which can create tremendous national debt for a sporting event that lasts 16 days.  Most of the budget-draining costs are centered in operation and construction during the four years prior to the event.  If the organizing committee believes that costs are going to exceed allocated funds, the committee can petition to the national government (parliament in particular for Norway) for increases in the budget. Some of these petitions are successful while others take money away from other Olympic projects.

The Norwegian Parliament produced five budget propositions between 1990 and 1995 for the '94 Winter Olympics in Lillehammer.  To obtain five different propositions, the budget was amended on four separate occasions by parliament.  The estimated final cost was 379,514 Norwegian kroner over the original budget from 1990 (roughly $50,000 USD).  Did any of these budget changes by the Norwegian Parliament predict exceeding the original budget more than other years? 

To analyze the relationship between the individual budget changes and the predictability of exceeding the original budget, I created a logistic regression model to compare coefficients between the years to establish relationship strength of the individual budget change to the eventual outcome, overspending.  I looked at Lillehammer's budget expenses (obtained in the Lillehammer Olympic Official Report) and selected 20 expenses that were covered in all five parliamentary propositions.  The chosen expenses range from sport venues to marketing costs to cultural venues, covering all major aspects of hosting the Winter Olympics.  The individual budgetary changes made by parliament serve as the predictor variables. The Norwegian Parliament made changes to the budget in 1991, 1992, 1993 and 1994. For each expense, I designated whether the budget allocation increased or did not increase (either staying the same or decreasing) in that year.         

The chart from my model and the results from the regression test are shown below:

[1 = increase; 0 = did not increase] 

Norwegian Parliament Budgetary Changes

Coefficients from Regression Model According to Year

The year 1992 had the highest value for a positive coefficient which indicates that increasing an individual expense in that year had the highest probability of going over budget.  Conversely, deciding not to increase an individual expense yielded a lower probability.  The year 1994 had the lowest coefficient as most expenses were already over budget and attempts to shrink the budget failed to make a difference in the eventual outcome OR projects were already sufficiently under-budget to the point that increasing the allocated funds made no difference in keeping the expenses below the original budget.

These coefficients explain that the budget changes in 1992 (in which 15 out of the 20 selected expenses were increased by the Norwegian Parliament) indicated the highest probability of going over the original budget.  The parliamentary decisions in 1992 had strong effects on future parliamentary decisions as opposed to 1991, when the construction process was still fairly new and unfamiliar, thereby making budget changes less predictive of future action.  The increased probability of exceeding the budget after 1993 is what prompted parliament to wisely cap the total Olympic budget at 7,374,514 kroner in 1994.  Perhaps if Russian officials had done the same, Sochi wouldn't have produced $2.5 billion in estimated national debt... 

Lillehammer Olympic Budget from Official Report

Linear Optimization: Maximizing Participation in Opening Ceremonies

When planning the Opening Ceremony for the Winter Olympics, the host country must balance artistic vision with the practicality of budget and time constraints.  The financial success of the Winter Olympics depends largely on the popularity of the Opening Ceremony.  A major percentage of the host's total revenue depends on television sponsorships; therefore, the Opening Ceremony is the perfect place to spark interest in the games, increase viewership and recognize the host nation's rich and unique culture.  An interesting Opening Ceremony will result in higher viewership for future programs.  Likewise, a lackluster performance may discourage viewers from watching the whole program or tuning in to future events.  A limited budget unfortunately constrains the amount of participants involved in the spectacle which can limit variety of cultural elements included to represent the host nation.  In 1994, Norway struggled with what to include in the performance, hiring and firing several producers and designers in the process.  The tumultuous process is documented in the third volume of the Lillehammer Official Olympic Report.  The LOOC agreed on having the following elements in the Opening Ceremony: Sami joik music, Telemark skiers, a children's choir, a traditional Norwegian wedding ceremony, fiddlers and folk dancers,

Lillehammer Opening Ceremony Postcard from 1994

After deciding on the different cultural elements, the LOOC had to decide on how many participants would be involved in each performance. They wanted to maximize the amount of participants while still maintaining a strict budget of approximately 60% of the Main Ceremonies expense given in the parliamentary propositions.  This limited the committee to approximately kr 57,000.

To determine how many participants would be involved in each performance, I created a linear optimization model and constrained the results underneath the budget limit.  I approximated the cost per participant in each performance, taking into account costume expense, prop expense and/or professional fees.  For instance, the children's choir has lower costs because the children were performing for free.  The traditional Norwegian wedding ceremony would be more expensive per performer because of the intricacy and expense of the costumes and accompanying horse-carriage needed for the ceremony.  For restraints, I required that there be at least 5 Sami joik musicians, 10 performers in the traditional wedding procession and 2 skiers to make sure those elements were clearly represented.  (Otherwise, ideally, the Opening Ceremony would consist of inexpensive, amateur children's choirs)  I also capped the participation of the children's choir at 250 because a larger group of children would require additional supervision, coaching and rehearsals which the hired producer would have to account for in his artistic vision.  Rehearsal time and location was also already an issue in planning for the ceremony and was highly contested among the different producers.

The optimal result was 5 Sami joik, 2 Telemark skiers, 250 children in the choir, 10 participants in the wedding ceremony and 170 fiddlers and folk dancers.  This result allowed 437 Norwegian citizens to participate in the Opening Ceremony at a total estimated cost of kr 56,810.

My Linear Optimization Model
*Note that data besides budget cap is completely arbitrary*

Network Model: Delegating Segments of the National Torch Relay

One of the oldest traditions surrounding the Winter Olympics is the Olympic Torch Relay.  The Olympic flame holds ceremonial importance and establishes a strong connection between the original games and the modern Olympics.  In 1994, the LOOC decided to add an additional relay to the ceremonial procession that focused on Norway specifically.  This additional relay was deemed the National Torch Relay and began in Morgedal, Sondre Norheim's birthplace.  Relay participants were average Norwegians and the relay lasted 75 days to induce nationalism.  Approximately 1 million viewers witnessed the National Torch Relay and participated in the various cultural activities and events accompanying the relay.

The Norwegian Post designed the route of the National Torch Relay, including 11 stops in 9 different cities and concluding in Lillehammer.  Once the route was designed, the Norwegian Post had to decide which Norwegians would cover which sections of the relay.  

In an effort to showcase my understanding of network modelings for my Excel modeling project, I created the following hypothetical situation: the Norwegian Post has selected 12 individuals as relay leaders with the responsibility of organizing participants to carry the National Torch from one city to another.  The 12 individuals were given a survey asking to rank the legs of the relay in order of personal preference, 1 being the most preferred leg and 12 being the least preferred leg.  Theoretically, the most preferred leg would be chosen because of interest in the route, knowledge of the area and connections with other individuals in the area.  I created a network model to optimize the delegation of relay portions to maximize the preferences of the 12 individuals in accordance to their survey submissions.  

This model was very useful in taking the survey data and assigning the participants to different legs of the National Torch Relay that were closest to their top preferences but still in the best interest of the group.  Solutions like the one obtained from the network model are helpful in maintaining a strong relationship between Olympic organizers and volunteers.

Survey Results

Network Model

Results

Ingrid- Bergen to Gullfaks

Lars- Oslo to Lillehammer

Hendrik- Svalbard to Tromsø

Ole Christian- Morgedal to Kristiansand

Singrid- Gullfaks to Bergen

Agnes- Bergen to Trondheim

Jacob- Stavanger to Bergen

Marta- Bodø to Oslo

Espen- Trondheim to Tromø

Kristian- Tromsø to Svalbard

Erik- Kristiansand to Stavanger

Kristoffer- Tromsø to Bodø

Integer Optimization: Selecting Lillehammer Art Museum Exhibits

The Lillehammer Art Museum hosted a total of five art exhibits during 1993, the year before the Winter Olympics.  These five exhibits were designed to recognize Norwegian sculptors, painters and artists, sharing their work with both domestic and international visitors. The earliest exhibit began November 1992 while the last exhibit ended January 1994.   A total of 79,667 individuals visited the Lillehammer Art Museum to view the five unique exhibits.

 If the five exhibits do not overlap, the Lillehammer Art Museum is reserved for these Olympic exhibits a total of 381 days.  Suppose the museum limits the exhibits' stay to a year (365 days).  The museum's limitation forces the LOOC to eliminate an exhibit in order to meet the new guidelines.  Assuming that both the LOOC and the museum want to maximize the number of visitors in 1993, which exhibit can be cut from the exhibit selection?

To solve this problem, I created an integer optimization model designed to maximize the number of visitors while accounting for the time constraint presented by the museum.  The number of days total must be less than or equal to 365 in consideration for the museum's limitation and is included as a constraint.  Another constraint was added to ensure either Exhibit C (featuring Munch) or Exhibit E (featuring internationally-acclaimed modern art) were included in the selection, assuming they would be two of the more popular exhibits.  The optimal solution recommended that Exhibit D (featuring international paintings, jewelry and ceramics) be eliminated from the selection.

More information on the individual exhibits found on page 129 of the third volume of the 1994 Official Lillehammer report.

Decision Analysis: Submitting a Bid

The decision whether to submit a bid for the Winter Olympics is a risk that many countries are unable to take.  For many nations, the costs associated with bidding outweigh the benefits of being chosen to host.  Beyond the cost, an important consideration is how national pride and the country's reputation will be affected by the IOC's bid eliminations and selections.  These effects are an important consideration for politicians and public officials when deciding whether or not to submit a bid.

In 1994, there were four locations that submitted bids to serve as the host city by the IOC.  The four were Lillehammer (Norway), Östersund (Sweden), Anchorage (USA) and Sofia (Bulgaria). Once host cities submit a bid, the IOC has three rounds, eliminating one location each round.  The selection process occurred on September 15, 1988 and the final results can be found here. 

To model the effect of the bidding process on the country's reputation to both the IOC and the international sport community (i.e., future consideration to host large events like the World Cup, Summer Olympics, etc), a decision tree model was designed to model the expected utility from each possible outcome.  Positive values signify improved reputation & public sentiment while negative values signify diminished reputation and public sentiment.  For example, an end utility of -50 after being eliminated from the first round in the bidding process illustrates negativity associated with failing to receive a bid, whether it is public opinion or future considerations regarding huge events.  An end utility of 185 after being chosen to host illustrates increased nationalism and international reputation.

Choosing not to submit a bid would preserve the nation's reputation and would slightly increase the chance of getting to host a huge event in the future.  For example, the Scandinavian countries had not hosted a Winter Olympics in a while which contributed to the probability of either Lillehammer or Östersund being selected for the 1994 games. When deciding whether or not to submit a bid, an important consideration is the likelihood of being chosen.  Bidding when the possibility of elimination is high is fiscally irresponsible and showcases countries' flaws (which are often highlighted during the bidding process).

The values below are arbitrary and based on their relation to other values in the tree (being eliminated in the first round is slightly better than being eliminated in the third round where more money is spent by the government without being chosen, disliked more by citizens).  Looking at the end utility values, it is evident that Norway had a lot to gain in nationalism and public sentiment if chosen as a host nation.  Whether Lillehammer or a bigger city like Oslo should have been chosen is still up for debate and will be later discussed in future analyses.