Category Archives: Demand Estimation


After reading about an application of how estimation/regression analysis can be used in forecasting, consider the following:

  • Is it odd that athletic-related determinants are rarely mentioned in this estimation?
  • How can empirical functions estimating Olympic medal winnings be used in future Olympic Games?
  • Can you foresee any pitfalls in using the result of past Olympic Games to say anything about the future?

Our classroom discussion will take place after midterm exam #1. I am looking forward to it!

Posted by Prof. C-S


Demand Estimation

Discussion on “China Goes for (All of) the Gold: Economists predict whether the host country will rule the Beijing Olympics” (Slate, 2008) and “China’s Winning Ways:  Did economists correctly predict who would win at the Beijing Olympics?” (Newsweek, 2008), both by Daniel Gross.

Summary & Discussion of Relevance

China Goes for (All of) the Gold

In this article, Gross compares different economists’ predictions on China’s and other top 10 countries’ results in the 2008 Beijing Olympics.  This “medal-count guessing game” took place between John Hawksworth of PriceWaterhouseCoopers and Andrew Bernard of Dartmouth’s Tuck School of Business.  The purpose of the forecasting grew from the question of how China’s recent expansion and influence in the competitive, global market would impact their Olympic performance, especially considering their home-field advantage in Beijing.

Hawksworth at PwC projected China to out-do the U.S.  Due to increased globalization, he forecasted that market share would shift from the industrialized West to the changing landscape of wealth and progress in expanding markets (China, Brazil, Indonesia, Mexico, Poland).  Bernard, on the other hand, predicted that wealthy nations would continue to flourish while the emerging world would take “less of a bite.”  However, both Hawksworth’s and Bernard’s shoddy predictions from the 2004 Olympics set the stage for equally miscalculated estimations in 2008.

Several reasons for why their projections were wrong were the fact that it may take decades for economic change to translate into dominant athletic achievement; cultures with athletic prowess, like Russia, can withstand economic downturns; sports weave in and out of culture, geography, and traditions that do not react to economic change; and predictions about the Olympics are difficult due to the nature of the events.

Gross argues that the best way to achieve Olympic success is by attracting, retaining, and developing human capital to its fullest potential, and that no country has done that better than the United States.

China’s Winning Ways

Gross wrote this article in response and as a follow up to his article about economic models and results forecasting written before the Beijing Olympics in 2008.  John Hawksworth of PwC and Andrew Bernard of Dartmouth’s Tuck School of Business compared factors such as home-field advantage, the size and growth of national economies, and former political affiliations.  After comparing the results, Hawksworth and Bernard underestimated both the sustaining power of the U.S. and China’s performance as host country.  Another surprising trend found that the wealthy remained wealthy, despite an economic shift towards a flatter world.

As previously mentioned, several factors that influenced the divergence between model and results include the home-country effect and economic growth.  Two other factors include a post-hosting letdown, where the previous Olympic host country (Greece) plummets in performance in the succeeding Games; and a pre-hosting boost, where the next country in line (Great Britain) begins to raise its own standards in preparation for its hosting duties.

Gross describes the difficulty in estimating results of a scenario like the Olympics with a top-down perspective.  By only basing estimation on past models, chance, randomness, and cultural phenomena cannot be considered.  He gives the example of how a nation’s openness to immigration may only a possible factor if economists consult experts in each sport and are able to pick potential winners.

These articles are closely connected to what we are focusing on in class.  As we finish up discussing the importance of consumer demand and utility, the articles demonstrate how shifts in the cultural, political, economic, and social geographies of certain areas of the world can have a significant (or insignificant) impact on predictability and forecasting.  These articles highlight the role that demand-side determinants play in estimation, as seen in the causes and effects of post-hosting letdowns or pre-hosting boosts.  Lastly, these articles show how differences in perspectives — regression models versus consulting experts in the sports — affect the success or failure of estimation.


The determinants of medal-winning in the Olympics were estimated by PricewaterhouseCoopers and Andrew Bernard of Dartmouth by considering the following: the general state of the countries’ economy based on indicators such as purchasing power parity; population; past performance in the Olympics; home-field advantage; state-sponsored sports programs; athletic culture; and human wealth. Past performance is the most concrete way to estimate medal numbers, and the estimations made were generally based on these past numbers. The history of communism and state-sponsored sports programs in countries such as China and Russia seem to have a lot to do with their past and continuing success.

As far as estimating changes in medal count between 2004 and 2008, both analysts believed that China’s booming economy and home-field advantage would lead to more medals. Indeed, China’s medal count rose by 59 percent to 100 medals including 51 gold.

The ability of these economists to predict the medal results with fairly good accuracy is impressive considering the nature of the games and how much chance alone can make the difference between a win and a loss for an athlete. A good run and a bad run for one athlete might mean the difference between a gold medal and no medal. Estimating home field advantage and other intangible influences is also very hard to do from a scientific standpoint. A more comprehensive way to estimate winners and losers might be to look at each sport and competition individually, although that would require much more data and inferences than this general, economy-based approach.

Real World Application: Jeremy Lin

Over the last two weeks, a name that has littered the covers of sports publications and TV shows is Jeremy Lin. Lin, a graduate of Harvard College, has risen from being a hardly-played substitute to a name known by all in under a month. While Lin’s meteoric rise to fame has been something of awe, it has also posed certain difficulties and constraints on jersey manufacturers in respect to demand estimation.

Like all other manufacturers, Jersey manufacturers have certain factors that constrain the amount of output they are able to produce, and most often, this is seen through the functioning capacity of machinery and the availability of resources. Given the limited amount of resources, a manufacturer would have to determine how best to allocate their resources in order to maximize profits. Specific to the Jersey manufacturing industry, this often means that manufactures will use determinants such as player popularity and team popularity to determine which and how many jerseys to produce.

However, when there is a sharp change in what is desired by consumers, Jersey manufacturers will often feel financial woe in terms of unrealized potential profits. In the case of Jeremy Lin, no Jersey manufacturer could have estimated that demand for the Jersey of a Knicks 2nd-string player would increase, but ultimately it did. As a result of Lin’s stellar performance, demand for his Jersey sharply rose, and the small supply of Jerseys was quickly depleted. If manufactures were able to somehow foresee this increase in demand and factor it into their demand estimation when choosing which products to make, they could have garnered a much greater deal of profits.

 In the case of Jeremy Lin and the shortage of his Jerseys, we see how unpredictability can muddle demand estimation and cause problems for producers.

Posted by Ben, Kyle and Kevin (Section 4)

Demand Estimation

Discussion on “China Goes for (All of) the Gold: Economists predict whether the host country will rule the Beijing Olympics” (Slate, 2008) and “China’s Winning Ways:  Did economists correctly predict who would win at the Beijing Olympics?” (Newsweek, 2008), both by Daniel Gross.

Top accounting firm PWC and renowned economist Andrew Bernard attempted to apply the economic principles of demand estimation to predict the number of medals each country in the Olympics would win. To create their models they used factors such as population and income levels, which are necessary for demand estimation of economic markets, but were irrelevant in determining how many medals each country would take home. They considered other factors in their predictions that were more relevant, however, to accurately determine medals wins using their models is nearly impossible.

The determinants they included were : homefield advantage, size and growth of national economies, past political affiliations, market share. The determinants had a correlation with winning potential, but did not necessarily indicate causation. They might have boosted chances but, by such an insignificant amount that it didn’t determine winning potential.  

Retail buying is another example of demand estimation. Buyers are responsible for stocking stores with the appropriate amounts of the right items. For example, buyers have to look at various factors in demography to determine what quantity of sizes of a t-shirt their company might sell in any given area.

Posted by Kelsey, Blaise and Crystal (Section 3)

Demand Estimation

Discussion on “China Goes for (All of) the Gold: Economists predict whether the host country will rule the Beijing Olympics” (Slate, 2008) and “China’s Winning Ways:  Did economists correctly predict who would win at the Beijing Olympics?” (Newsweek, 2008), both by Daniel Gross.

In Daniel Gross’s first article “China Goes For (All of) The Gold”, Daniel Gross analyzes PricewaterhouseCoopers and Andrew Bernard’s (Tuck School of Business) methods for projecting total number of Olympic medals, and more specifically, gold medals won by certain countries in the 2008 Beijing Games.

These methods, however, do not look at variables directly affecting athletic performance. Instead, PricewaterhouseCoopers and Bernard project Olympic success based on population, income level, economic status, past performance, home country advantage, and prior membership of a Soviet/Communist bloc.  Based on these criteria, PwC believed that much of the statistics played in China’s favor, and thus projected an overall medal increase of about 40%, from 63 to 88, and into the world leader for the 2008 Olympics. PwC also projected that countries such as Mexico and Brazil would see noticeable success, while America, Germany, Russia, and France would suffer a setback.  Bernard, on the other hand, saw “the rich getting richer,” placing the US in first, Russia in second, and China in third.

After outlining methods and projections for both PwC and Andrew Bernard, Daniel Gross then discusses his skepticism of both methods. He believed that both projection models overestimated the US decline in worldwide standing, home country advantage, and the loss of market share of Western powers. Gross believed that neither PricewaterhouseCoopers nor Andrew Bernard could accurately project Olympic success without considering more intangible variables, such as the cultural value of sport in different countries. Also, the Olympic results hinge on many other things that can happen during the games, unrelated to economics, like sudden injury to an athlete, or other unpredictable events during the games.  Ultimately, Daniel Gross argued that Olympic success is determined by a nation’s ability to recruit, develop, and retain the best athletes available to them, not economic factors.

In his follow up article, “China’s Winning Ways”, Gross discusses the mix of successes and failures of the PwC and Bernard economic projections.  One of the biggest successes was Bernard’s ability to predict the continued success of the US by correctly predicting 105 medals won, most in the world. PwC, on the other hand, severely underestimated the performance of the Americans.  The biggest failure in the projections involved the underestimation of home country advantage and the performance of China, who finished second in the medal count, and had the most gold medals.

In his conclusion, Gross determines that although a purely economic model would have helped predict China’s success, the variability of cultures in different counties regarding the Olympics make it hard to find a consistent model. For the most part, “the rich generally stayed rich.”

The two articles, mention a multitude of determinants that could affect a country’s medal count in the 2008 Olympics. These include “the basic assumptions that population (more potential competitors) and income levels (more resources to develop competitors) are crucial determinants of Olympic success” as well as “past performance, having been a member of the Soviet/Communist bloc, and/or home field advantage” (Gross 1). While computing the effects of the first group of determinants is easily quantifiable, creating models, such as regressions, to calculate the effects of the later groups proves more difficult. These determinants are more qualitative in nature, and thus make estimations difficult.

Furthermore, the magnitude of the effects of the determinants proves to be rather subjective. While both of the economists performing these tests, PwC and Bernard agree upon the determinants, they do not necessarily concur on how much of an effect these factors have in influencing a country’s medal count. For example, PwC places a greater emphasis on home advantage, while Bernard stresses the importance of a country’s economic prosperity in winning medals. The magnitudes placed upon the determinants thus prove to be highly subjective. This creates a lack of uniformity amongst estimation models and their results.

Finally, the determinants used by the economists disregard the human element in competition. While factors such as past performance and national wealth provide neat medal estimations, they do not include the key determinant of human error. A defining feature of the Olympics is the crushing upset when the projected victor loses his or her competition due to an injury, mistake, or the like. The models do not include this crucial factor in their estimations. Similarly, they do not account for surprise victories where a previously unnoticed athlete shockingly wins his or her event due to adrenaline, unrealized skill, etc. Because of this disregard, the qualitative nature of determinants, and the subjectivity of the determinants magnitudes, forming accurate estimations proves problematic.

The regression idea that the economists used to estimate Olympic Medals in Beijing is akin to how companies conduct demand estimation. Demand estimation is exactly what it sounds like; it is the estimation of the demand of a good or service through the use of various techniques, but mainly statistical regression.

One real world business application of demand estimation is the estimating the demand before the launch of a new product. Take for example, the launch of the iPad. Apple’s industry shifting tablet debuted in 2010, and became almost immediately unavailable at both Apple stores and retail outlets. Apple, and analysts as well, failed to correctly estimate demand for their tablet and thus vastly under-supplied retailers. The following table shows the availability of products at one of Apple’s main retailers Best Buy; it is important to note that the iPad was launched in mid April, and was still sold out in June.

As illustrated in the table (see link above), iPad demand was sufficiently under estimated. The iPad was continuously out of stock in retail stores, and Apple lost a great opportunity to maximize their profits. Even with the iPad 2, demand was originally underestimated by 40% according to analysts from Morgan Stanley. Apple could maximize their profits and make even more money by making better demand estimations. Apple could do this by taking in trends and factors such as price, income, ease of use, general interest, and the market for tablets to better formulate a regression model for demand estimations.

 Posted by Matthew, Beth and Jack (Section 2)


Demand Estimation

Discussion on “China Goes for (All of) the Gold: Economists predict whether the host country will rule the Beijing Olympics” (Slate, 2008) and “China’s Winning Ways:  Did economists correctly predict who would win at the Beijing Olympics?” (Newsweek, 2008), both by Daniel Gross.

Demand estimation is a way by which managers make economic or policy decisions based on estimates of the market’s demand function. The two articles give an account of how economists have tried to predict the number of medals that China would win during the Beijing Olympics in 2008.

The first article, “China Goes for (All of) the Gold: Economists predict whether the host country will rule the Beijing Olympics,” discusses how PwC and Andrew Bernard of Darthmouth’s Tuck School of Business have used different models to forecast the number of medals to be won. Both models are based on assumptions that first, population and income levels are crucial determinants of success in the Olympics, and second, other factors like past performance, past political affiliations, and/or home-field advantage enable countries to perform better. PwC projected that China will increase its medal count by 40% due to the home country effect and the state support for sports. Bernard predicted that China would only make modest gains, as he sees the “rich getting richer.”

The second article, “China’s Winning Ways:  Did economists correctly predict who would win at the Beijing Olympics?” critiques how economists have done in their projections. It reports that both analysts underestimated the strength of the US delegation as well as China’s performance. Overall, it was seen that the combination of nationalism and rapid economic growth had a greater impact on China’s medal count than expected. In addition, it found that despite rapid economic growth in emerging countries and increasing decentralization of the global economy, the rich generally remained rich. The results proved that predicting Olympics outcome is very difficult especially because chance and randomness play a big role as well.

Both articles were very relevant to our classroom topic on demand estimation because it listed out the dependent and independent variables for estimating the demand. In this case, the independent variables were factors such as GDP, economic growth, past performance, past political affiliations, and home-field advantage, while the dependent variable was the number of medals won. The variables can be put in a demand function based on their magnitude of impact and whether it will increase or decrease overall demand.

Even though GDP as well as home-field advantage provides a relatively good estimation of the medals won in 2008 Olympics, there are problems with this approach. The first problem is that the current GDP might not have a direct influence on athletic performances.  But for countries that had rapid economic growth in recent years, like Brazil and India, it will generally take years for GDP to translate into better sports facility and even longer for a culture that produces great athletes. The second major problem is that it overlooks the difference in culture with regard to Olympic sports.  Countries in former Soviet blocs have a special cultural and social structure that can produce better athletes which allows them to win more gold medals than countries that have similar GDP.  The third problem is that the impact of home-field advantage is really difficult to estimate, because it is different for different sports. The home-field advantage in sports like basketball may be great and for sports like shooting it would not be that huge.    

A real-world application of demand estimation would be high school graduates and the chance that they apply to Notre Dame.  There are many independent factors that way in on their decision to apply.  These include high school GPA, SAT and ACT scores, resume strength, legacy parents, home location, and preferences.
The first three variables determine if the student has the chance to get in to a school of the quality of Notre Dame.  The university takes only the smartest of high school graduates, and students can prove their worth by these measures.  They will know by these academic standards whether or not they have a chance to get in, or whether it would not be worth the time.  The next three variables have more to do with why students would choose to apply to Notre Dame over similar schools.  

Students whose parents attended Notre Dame are much more likely to want to go to Notre Dame, as they probably grew up fans of the school and feel influenced by their parents’ love of the university.  Where the student lives is also an important variable, as students who live in cities with strong alumni networks are exposed to the university more than ones who do not.  Cities like Chicago, New York, Los Angeles and Denver, which have huge Notre Dame alumni networks, tend to send numerous students across the country to northern Indiana to attend Notre Dame.  Preferences of the student also play a large role.  What this means is that a very academically minded person who is picking between Northwestern and Notre Dame may choose Northwestern, as it is slightly higher ranked in the university rankings.  However a student who values athletics and tradition may choose Notre Dame instead, as it offers one of the greatest football programs in history as well as many campus traditions that have been around for decades.  

Posted by Ava, Michael and Tiandong (Section 1)