Markets vs Polls

The march issue of Scientific American published an article: When Markets Beat the Polls.

The first thing I want to say is that I'm glad public opinion polls are considered "scientific" and discussed in a periodical related mainly to pure science and technology.

The article, however, is not about polls per say after I read it. Polls were used as a contrast to the main topic: market analysis.

The concept of the stock market was applied to predict election by a group of University of Iowa scholars in 1988 when Bush senior and Dukakis were vying for the presidency. The researchers set up an Internet-based interface so that people (traders) can buy "contracts" on Bush or Dukakis. It functions like a stock market where people buy or sell their stocks depending on how they expect the stocks to be in the future. So, for example, if a Bush contract costs you $.53 to buy it, it may signal that Bush would obtain 53% of the votes on the election day.

Although the characteristics of the traders are not nationally representative, the Electronic Market often predicted better than public opinion polls, according to some data in the article. The "free market" and "the wisdom of crowds" ideas are definitely fascinating. But its assumption is somewhat contradictory to my recent understanding of voting behavior.

First, economists have believed that people will make "rational" decisions, which means people will weight risks and benefits, and make a decision that maximize their advantage. However, in the field of public opinion and election, it has been found that people are not always rational--they don't collect every bit of information they need when deciding whom to vote for. In other words, the public is not information "maximizer". They are information "optimizer," those who use whatever information is accessible at the time of decision making. There are even occasions where people don't need information at all.

As Popkin illustrated in his seminal book "The Reasoning Voter: Communication and Persuasion in Presidential Campaigns," voters sometimes reply on heuristics or information shortcuts when making voting choices. These shortcuts save people a lot of cognitive energy that would have been devoted to rummaging through different pieces of information. Some examples of heuristics include party identification, religious beliefs, and people's ideological preferences.

So, if people are not rational most of the time, how can a mechanism based on rational decision making function so well? Well, I don't know.

The second point is about how close the outcome is related to an individual. In an election, many people couldn't care less about who get elected because they don't think their life would be affected. However, in some of the electronic markets, traders are investing real money. The final outcome will affect how much they earn or lose monetarily. Therefore, how people make decisions in these two different scenarios is interesting for further investigation.

But this is definitely an intriguing topics especially when it provides a different way to measure public opinion. Unfortunately, there is no free electronic file of the article to share with those interested.