Artificial Markets for Determining Energy Future

What they are

Artificial markets are online games based on the stock market, applied to the prediction of future events. Players buy and sell shares at prices determined by the market. Examples of stocks include, “Yahoo! Will outperform both Google and Microsoft Today,” “XM will have more than 5.5 million subscribers by the end of 2005,” “Carrie Underwood will win American Idol,” and “How well will House of Wax do at the box office?”

Artificial markets mainly fall into three categories: Those that involve exchange of real money, those that involve exchange of play money strictly as a game, and those that involve exchange of play money with cash prizes awarded to the winners. A real-money market is the Iowa Electronic Markets, operated by the University of Iowa on a not-for-profit basis for the purposes of research and education. The amount of money an investor can put into these types of markets is typically limited to a few hundred dollars. A purely play-money market is the Hollywood Stock Exchange, where players bid on movie box office performance, Oscar winners, American Idol winners, etc. Two examples of markets with cash prizes are Newsfutures.com and MIT Technology Review’s Innovation Futures. Interest in artificial markets has increased as recent studies have suggested that they can predict outcomes of future events as well as experts can. The reasons are that well-informed investors tend to be rewarded in the market, while those who rely on luck or try to artificially inflate the value of a preferred outcome tend to do poorly. Play-money games have been shown to be comparable to real-money games in accuracy.

How they work

Players buy and sell shares based on propositions for future events. Shares are initially introduced at a neutral price. They are then allowed to trade at market prices until the event occurs, at which time they are cashed out at a value determined by the outcome. A variety of schemes exists for determining cash-out value. In the Hollywood Stock Exchange, for example, movie stocks are awarded points based on the box office gross after four weeks. American Idol winners are awarded H$1 for the first of the 12 finalists to be voted out, H$2 for the second, and so on up to H$12 for the winner. An actor up for an Oscar is worth H$25 if he or she wins and H$0 if he or she loses. (The “H” prefix indicates that the amounts are Hollywood Stock Exchange play money and not real money.)

Players can gain or lose either by holding on to shares until they are cashed out, or by buying and selling shares with other players before the cash-out time.

Application to Medvick’s Matrix

Richard J. Medvick has done an excellent job in creating his matrix on energy sources and usage, and the plan to populate the cells with rankings of the relative values of each technology is a great idea. An artificial market based on this matrix could possibly be an economical way to obtain values for these cells. Results obtained by this method may be as accurate as expert judgments. One caveat is that the markets that have been used for research study have typically focused on short-term events, ranging anywhere from one day to one year. In addition, it is generally possible to cash out the shares at some future date, meaning that an outcome can be determined on that date. That is different from the proposed matrix in which we hope to make predictions 5, 10, 20, and even 50 years out. A possible problem that I foresee with this is that the market could stagnate and people might lose interest unless there are some short-term rewards for staying involved.

Details of the implementation of this method would require further study in order to determine the best way to approach it. I would expect that each cell in the matrix would be the basis for one type of stock. One approach would be to base everything on true/false premises, which seems to be the most common approach. Examples might be “Dark matter will become a viable energy source by 2010,” “Biofuels will reach $1 billion in sales by 2010,” or “Fuel-cell-powered car sales will exceed one million cars by 2015.” Another approach that is less often used is to base the value on actual sales. Examples might include, “Total sales of biodiesel in 2010,” “Total sales of dark matter in 2050,” or “Total sales of hydroelectric power in 2020.” At the end of the given year, the cash-out value would be a percentage of the actual sales for that year.

At the last energy roundtable meeting, it was suggested that the proposed method for filling in the cells, using experts to research and make predictions about each technology, be tried on a small scale using a portion of the entire table. An artificial market could probably be implemented for the entire matrix at a much lower cost. The small-scale study using experts could be used to validate the results of the artificial market by comparing results among those parts of the matrix that are evaluated both ways.

Clifford F. Lewis Consultant Center for Regional Economic Issues Case Western Reserve University 10900 Euclid Avenue Cleveland, OH 44106-7235 (216) 371-3981 Fax: (216) 368-5542 innovate@case.edu

Return to May 6, 2005