Prediction Markets in Science

In a short piece in the latest Science journal, about the Promise of Prediction Markets, we provide a short review of the literature on prediction markets — how and why they work, and the accumulated empirical evidence.

But our key point is public policy:

Unfortunately, however, current federal and state laws limiting gambling create significant barriers to the establishment of vibrant, liquid prediction markets in the United States. We believe that regulators should lower these barriers by creating a legal safe harbor for specified types of small stakes markets, stimulating innovation in both their design and their use.

The rest of the piece goes into greater detail, suggesting a way for the C.F.T.C. to provide a safe harbor for research-based prediction markets, and for companies or government agencies to set up small-scale internal prediction markets.

I am one of 22 co-authors of the piece, although the real credit goes to Robert Hahn who has been pushing hard on these issues.

The fun part for me is that I can now claim three Nobel Laureates as co-authors: Ken Arrow, Thomas Schelling, and Vernon Smith.

I was a bit surprised that this illustrious list of 21 co-authors didn’t lower my Stiglitz number, but then remembered that Joe is a prediction market skeptic (despite his earlier scientific writings on futures markets).


Just curious . . .valuable though they may be, what, specifically, makes prediction markets "not gambling"?


The very utility you mention is what makes them not gambling. They can serve a purpose, especially for people who would be significantly affected by the result. If I am planning to move out of the country if another Republican becomes president, I might want to hedge the risk of my moving expenses, for example. Obviously most people won't use them that way, but the information they provide is also "utility," and there is good evidence that we can't get that information as efficiently through other means.

The utility argument is used in many other cases of contracts involving, on their face, much more risk than you'd find at Vegas (such as stock options, commodity futures or insurance).

Of course, laws against gambling are already completely arbitrary and silly, so why should this be any different? After all, if people can "bet" on future events, they won't pay as much lottery tax to the government. Good luck with that.




There are two arguments for prediction markets not being gambling.

The first is, if you're willing to call them gambling, are you also willing to call any sort of risky investment gambling? (Stocks, bonds...)

The second is, there's a lot of skill to being a successful trader if you don't want to lose your shirt*. Yes, there's some randomness at play, but I think if you compensate for the biases of the media and traders, you can eliminate a lot. (Ron Paul and Al gore were both obviously overpriced, McCain and Obama were too discounted, and so on. I did make money on IEM trading these obvious analyses; I don't feel they were gambles.)

* You can't just bet randomly, since prediction markets aren't perfectly accurate. For example, if you were randomly betting on Intrade since the beginning of the year, you would've bought a bunch of way overpriced Ron Paul shares, for example, and not enough favorites.



Does it concern you at all that the futures markets are contributing a great deal to the current gas prices and subsequent economic domino effects? Yes, eventually gas prices will come back down when the crude oil market is corrected, but in the meantime a great many people are suffering very real harm for no other reason than speculation markets, while the underlying commodity has seen a much more modest change in supply and demand.

Projecting this to other prediction markets -- if, for example, political speculation markets existed on a large scale with significant money behind them, we would doubtless see very real cases of the market itself manipulating the political process it is ostensibly a mere reflection of.

Is it really such a good idea to encourage such needless manipulation of already precarious balances? After all, the prediction markets in political futures, for example, would provide no real utility to anyone other than academics and the people investing in them.

In almost every aspect of life where a prediction market provides genuine utility, they already exist in the form of insurance.



Trading in a prediction market isn't gambling because it is a market. Entering the market, you purchase a commitment (contract?) of $100 payment on the occurrence of an underlying event (an asset). One may resell the commitment when it is believed to be over priced. The prices are thus free-floating and set by the supply and demand of assets in a market.


I don't think anyone has persuasively argued against the question raised in Comment #1.

How is betting on the identity of the next American president different than betting on who will win the next Super Bowl? Does the former take more skill than the latter?

In a prediction market, contract prices are set by the "market" and you are betting against other participants--Intrade is simply the broker. This is exactly how betting works in a horse race--the more money that goes to a favorite, the shorter the odds become.


What's wrong with gambling? Admittedly some people have gambling problems, but everyone is susceptible to some addiction (even if the addiction is to feeling superior to the addicted).
Other posters have commented that there is some special value to prediction markets, but that seems to be a specious argument. Why is predicting that Republicans will win the White House better than predicting that you will get a 7 of hearts to fill your inside straight? If we believe that free markets value goods appropriately, then clearly gamblers value the gambling buzz to the extent that they are willing to spend the money they spend on it.
Sorry, guys, exceptions for prediction markets don't cut it; let's just legalize gambling.

Christopher Smith

As an answer to comment #1: The difference between this and gambling is that what's effectively being bought and sold is information. There's a premium paid (the contract profit) for providing correct information, and if a "large enough" market exists, there is an incentive to develop skilled analysis tools.

To commenter #2, although speculators are the current favorite scapegoat, there's plenty of evidence that oil prices are staying high because of fundamentals--after all, if there were no fundamental reason that prices were high, consumption would drop and prices would readjust. Instead, we see a fairly small drop in the amount of oil demanded, and overall global demand is constantly rising (in addition, of course, to a weak dollar).