The idea of the market as ‘information aggregator’ is, like many ideas, probably as old as humanity itself, but Friedrich Hayek is usually credited with popularising the idea in his 1945 article “The Use of Knowledge in Society”. He writes
The peculiar character of the problem of a rational economic order is determined precisely by the fact that the knowledge of the circumstances of which we must make use never exists in concentrated or integrated form but solely as the dispersed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess.
The economic problem of society is thus not merely a problem of how to allocate “given” resources—if “given” is taken to mean given to a single mind which deliberately solves the problem set by these “data.”
It is rather a problem of how to secure the best use of resources known to any of the members of society, for ends whose relative importance only these individuals know.
While I agree that information aggregation is a problem in maintaining a ‘rational economic order’, I disagree with the religious fervour that ‘markets’ are promoted the ‘solution’ to the problem of information aggregation. This is particularly the case with prediction markets, which are often seen as a type of mystical seer of the future. But if markets have an ability to aggregate information, there must exists alternative processes for information sharing, since markets can only reflect existing information and not provide new information.
A common modern view is that prediction markets are so good at information provision, or prediction, they should be widely extended to more areas of life. Here’s just one framing of the issue
Economists believe that financial markets do a good job in aggregating information in part because they provide the participants with strong incentives to form good predictions.
Speculative markets perform relatively well when compared with information institutions (academia, news media, experts, etc) in terms of their information aggregation and prediction accuracy when presented with the same situation or environment.
Some examples may include Florida Orange juice commodity futures which have improved on government weather forecasts (Roll, 1984), betting markets that have beaten major national opinion polls 451 out of 596 times in predicting U.S presidential election results (Forsythe, Nelson, Neumann, and Wright, 1992), and betting markets that beat out Hewlett Packard official forecasts 6 times out of 8 at predicting the computer corporation’s printer sales (Chen & Plott, 1998; Plott, 2000).
So why not incorporate such markets in areas of public choice in order to facilitate societal decision making?
To really consider this topic we must first be clear about the economic and social value of information, what we mean by aggregate, and what ‘markets’, especially prediction markets, actually do. Then we can consider when and where markets seem to be useful tools for information provision, and when and where they probably are of little value. The intention being to wrap some social and economic context around an otherwise bare, idealistic, and often skewed view of the role of markets in society.
First then to the economic and social value of information. Say, as Hayek fairly rightly assumes, that information about how to produce certain goods, where to find certain resources and so forth is known only very narrowly amongst specialists. It is almost self-evident that this knowledge is of value to production, since both having the capital equipment and knowing how to use it are inseparable ‘factors of production’.
The question then arises, what information is needed by others from these specialists in order to facilitate production and trade? Surely not everyone needs to know how to build a car, grow grain, or raise cows. The specialist need only share that information with interested parties in those markets who typically pay for such expertise. Such sharing of expertise can be provided through labour markets and other contracts directly with the individuals who hold the knowledge, but it can’t be provided by prediction markets expect under particular circumstances.
Prediction markets merely reflect the marginal participants’ willingness to gamble on a future event. In a world where the density of predictions is very high and also fairly independent (most people are predicting much the same result, they not relying on the prediction of others to form their own predictions, and the marginal gambler is somewhere near the average), we get a pretty good signal from betting or futures markets about price expectations.
Of course there are other ways to aggregate information, and many ways to improve the available information without resorting to the provision of a common signal based on the beliefs of the marginal gambler.
For example, we have some evidence that commodity futures are correlated well seasonal weather patterns. But there is no new information in the futures price that existed prior the the market being formed – it merely reflects one method for aggregating existing information held by only those participating in that market.
Society as a whole would be better off with new information, and that only comes through investment in weather monitoring and forecasting institutions and capital equipment. The supply of such is typically seen to have characteristics of a public good and is typically publicly funded.
To make my case more clearly, betting markets only provide useful information if there already exists useful information to aggregate. Imagine if we we sitting around in the 1700s and trying to figure out the depth of the Pacific Ocean. We could open betting markets, but at some point we would actually have to measure the depth to actually produce new information rather than an aggregate of beliefs.
When a small number of market participants have differential abilities to control the future outcomes of the bet there are also serious problems with prediction markets. A recent survey article on the now long history of experimental testing of prediction markets say the following
Together these papers suggest that prediction markets are not universally the best choice. Rather, it is important to identify when prediction markets are a good choice. The answer depends at least in part on the extent of the robustness of these markets to insiders or manipulators.
This paper also summarised the experimental condition where there exists an ability for insiders to manipulate prices under a variety of conditions. Intrade, who extended betting to obscure outcomes, provides lessons about the ability for insiders to profit from their knowledge.
Will weapons of mass destruction be found in Iraq? Will Israel bomb Iran? The answers could alter global economics and politics—and Intrade ventured them, sometimes appearing to reflect insider information in the process. On December 13, 2003, U.S. military forces discovered Saddam Hussein in his spider hole in Tikrit. Up to that week, the Intrade contract for his capture had traded at a dismal 40 cents, indicating merely a four percent chance of his capture by the end of the year. Then, in the days before his capture became public, the contract registered unusual trade volume—driven, one can only assume, by someone who knew what was about to happen. By the time Paul Bremer said “We got him” to reporters, Intraders already knew. Anyone who bought a Saddam contract a week before had made a profit of 2,500 percent.
Do we really want to develop a whole espionage industry devoted to improving information for gamblers in prediction markets? As if there isn’t enough of this already in financial markets where actual ownership claims of tangible capital assets are at stake.
The real problem with wide adoption of prediction markets as policy guide is that incentive for those with money on the line to manipulate the outcomes. Politicians could bet anonymously on policy outcomes already decided behind closed doors. Powerful lobby groups can manipulate outcomes and gamble on their inside information. Imagine controlling insider trading is such broad prediction markets. I would suggest that prevalence of insider trading is one reason such markets have not taken off in their own right – only those with inside information would be willing to place bets, and no one would bet on the alternative outcome, given the knowledge that only insiders play these markets.
As the survey paper suggests, many true believers in prediction markets have suggested they be used as policy inputs is a wide variety of areas, from education funding to anti-terrorism. Of course the social question is whether the risk of having your intelligence agency relying on a signal so easily manipulated by the enemy is a wise choice.
We also know that surveys and opinion polling are pretty good aggregators of information as well. So why the devout attachment by the rational expectations crowd to gambling markets?
In almost every case where ‘prediction markets’ have been proposed, there exist alternatives that offer both new information a much richer set of aggregate information than single bets on prices and timing of future events.
For example, it has been proposed to me in conversation that tertiary education choice could benefit from signals about future earning by opening gambling markets on future earning of individuals studying different degrees and entering different professions. Student would then be able to use that signal to make a better decision about the choice of study and career path.
An almost costless alternative is to make publicly available individual income tax information along with main occupation. Say like Norway. Surely this would be more valuable information for a number of reasons, not just a guide to income of different occupations to aid career choices.
Other big calls are for wide implementation of ‘prediction markets’ as policy guides.
why not let citizens bet on, rather than submit to professional opinion on, for example, which tax policy is more likely to bring prosperity?
Yeah, maybe think that one through a little more.
The moral here is that markets don’t really aggregate information as is commonly believed – they in fact reveal the marginal beliefs of the participants in the market, or in the case of active manipulation of the market, absolutely nothing at all. All information that determines the prices already exists, and none of it is shared through prediction markets – only a single piece of information, the marginal price belief – is actually generated.