Election Prediction Markets

For reasons that escape me, several writers have recently taken to trashing election markets as good predictors.

Before I address their specific points, let me summarize my own view on election markets.

• If the markets are thick (have lots of potential buyers and sellers), they are likely to incorporate all the information available to the participants at any given point in time.

As Felix Salmon points out,

...  in the final hours or days before a big political result/announcement, prediction markets become less, not more, reliable.

For examples, Salmon relies on the recent pre-announcement contracts on Palin and the selection to be McCain's running mate, where the price jumped all over the place during the last 24 hours as new information hit the wires. He also refers to the New Hampshire Democratic primary.

Let me add that on the evening of the 2004 US presidential election, the contract price on a Bush victory jumped around considerably as the results of different exit polls were announced and revised.

Unfortunately, Salmon then adds,

There's simply no real-world justification for the degree of volatility we saw in the Palin contract this morning; this is not a consolidation of "the wisdom of those who traded in the past" but rather a crazed market resembling nothing so much as a bunch of six-year-olds chasing a soccer ball.

Salmon is over-stating his case. In all these examples, including the Palin veep market, more information led the market participants to change their expectations about the outcome, and this change was reflected very quickly in the market price. At any given point, the market was incorporating all the latest information into the predicted outcome. That is exactly what one would expect from an efficient market.

All this last-minute variability seems to make some people think prediction markets are silly, even though prediction markets (if they are thick) are notoriously better than polls at predicting outcomes of elections, especially a week or two before an election or announcement.

But what about the Canadian prediction markets?

Despite their questions about prediction markets in general, both Stephen Gordon and Mike Moffat point out that the Canadian markets have not been great predictors in the past. Their arguments hinge on one key observation: the Canadian prediction markets are extremely thin, or as MM says, "illiquid".

They both point to arbitrage opportunities on which they capitalized in the last Canadian federal election, and in his comments MM pointed out that there was a bias among the bettors in the market which he was able to exploit.

But if the Canadian election prediction market continues to exist and grows during the next election, these opportunities will probably disappear. Others, learning through Stephen's and Mike's blogs if nowhere else, will also see these opportunities, enter the markets, and drive the profit opportunities to zero. That is what has happened in sports betting markets, and it will happen in the Canadian prediction market as this market thickens.

On my own, I haven't bet on an election outcome ever before in my life. But when my friend and colleague, Salim Mansur, opined that John McCain is sure to win the US presidential election, I offered to bet him $50 on the outcome. He readily accepted the bet.

If I were truly risk averse and if transaction costs were zero, I'd hedge this bet in the Intrade market, where, as of this writing, an Obama contract sells for $60.

Comments

[Tom Hanna]

The moves of prediction markets on the McCain VP pick are comparable to the stock market's treatment of WorldCom and Enron before their problems became public. The company balance sheets showed that they were worth their market values or more. The McCain campaign's false leaks and misleading moves took Palin off the market entirely for the last week. I wouldn't be surprised if insiders were buying Pawlenty and Romney contracts to help throw off the media.