Prediction markets for the flu

In many asset markets the very phenomenon of market trading appears to increase volatility. Richard Roll did some empirical studies of exogenously determined market closings (occasionally various exchanges have closed for brief periods for purely technical reasons, such as catching up on paperwork or upgrading systems). Price volatility was higher across periods when markets where open than when markets where shut. In essence you draw inferences from watching other people trade, other people draw inferences from your inferences, and so on. The “dialogue” embedded in the market makes the price bounce around, even when it hones in on a better mean forecast.

Note further that most meta-rational people should not trade (economists refer to no-trade theorems). When you trade, you should always ask why you think you know more than the person you are trading with. Not every trader in the market as a whole can meta-rationally conclude that she knows more than the other traders. Yes you might trade for liquidity to put your kids through college, but I suspect that most trading in prediction markets is or would be opinion-based. It is especially prone to hubris and irrationality, which again contributes to volatility.

Here is the paradox. We need volume for the price to tell of much about market conditions. But volume is based on irrationality to large extent. Most people — the uninformed — should simply park their money about forget about active trading. And the remaining informed cannot so easily trade against each other.

Prediction markets may be least well-suited for predicting terrorist attacks. Tracking an excessively volatile price could create worry and perhaps sometimes panic. What will happen each time the market price spikes for a nuclear bomb going off in a major American city?

The very virtue of prediction markets now becomes their cost. If you hear rumors, in the absence of prediction markets, you can ignore them and pretend they are not true. With asset markets, however, your forecast moves into equality with that of the market, otherwise you would trade. It is precisely this “forcing quality” that makes prediction markets so useful, but also so potent. Price movements are materially and psychologically harder to ignore. The very feature of prediction markets that mobilizes information also mobilizes coordinated social reactions to the embodied information, and not always for the better.

So the prediction market skeptics have a valid point in some contexts, but this does not detract from the benefits of prediction markets more generally.


Comments for this post are closed