Wednesday, 5 November 2008

The US Presidential Election Results – what did the markets predict?

It is a truism often stated that what seems obvious with hindsight looked less obvious before the fact and this is no less so in the realm of finance than politics. And so as we all start to debate the inevitability of President-elect Obama’s victory it is worth reminding ourselves to what extent this result was in doubt as we approached the election date. Prediction markets are a useful device to check the extent to which we dabble in ex post rationalisation. They allow agents to trade the likely outcome of economic and political events in real-time in advance of and all the way up to the event itself.

The charts here show one well known prediction market run by Iowa University ( There were two US Presidential markets: the first for the respective share of Democratic and Republican votes and the second a winner-take-all market for the Party that would secure the greater number of votes. The vote share market represents the purchase or sale of futures on the Democratic or Republican share of the popular vote. On expiry, after the election, the future will be worth the fraction of US$1 equivalent to this percentage share. So, for example, early indications suggest that the Democratic share of the popular vote was 52% yesterday and so the price of the Democratic contract on expiry will be likely to be around 52 cents - meaning that anyone who bought at less than 52 cents or sold at more than 52 cents, prior to expiry, will make money. The official source for the liquidation price will be this Friday’s New York Times.

The first chart shows the prices of Republican and Democratic contracts since the contract started trading in June 2006 and we can see that apart from some glitches in May this year, there seems to have been a consistent lead for the Democrats, with some evidence of an increasing divergence since mid-September, after the collapse of Lehman Brothers. I do not have data on previous elections to hand but the consistency in the lead for the Democrats is interesting but so perhaps is the relatively small difference projected in the share of the popular vote. The spread of prices on the close Monday 3rd November data was 53.4-54.1 cents for the Democrats and 46.1-47.0 cents for the Republicans, which seems to have turned out to be reasonably accurate.

The second market is a simple horse race bet on which party will win the greater share of the popular vote with the all the pot distributed to those betting on the winner. In this case, the winner-take-all future is bought on either of the two parties at a discount to $1 with the sum of the prices equal to one dollar. The payoff on the contract, to those holding a future on the winning party, is paid for by the losses of those holding contracts in the losing party. Again the liquidation prices will be set by the report of the election in this Friday’s New York Times. So let us suppose that there was a 50:50 chance ex ante of either party winning the major share of the vote, this would be reflected in an equal amount of money being placed on either possibility and the prices of the two futures would trade at 50 cents. If this market felt that the Democrats became more likely to win, the price of the Democrat future would rise and that of the Republican future would fall. And in this case, as the Democrats have received the majority share of the vote anyone holding a contract for the Democrats, for which the price on Monday 3rd November close was 90.3 cents, will be paid $1 and anyone holding a Republican contract for which the price on Monday was just under 10c will get nothing and lose their stake.

The second chart shows a similar picture to the first, in terms of a consistent lead for the Democrats but also suggests that after the re-emergence of financial turmoil in September betting on the blue corner took a hold. With prices of the futures going from 53c on September 12th to over 90 cents by Monday morning, representing a 70% return in under seven weeks. Clearly Democratic Party Futures were the hedge against stock market turmoil. The prediction markets do seem to tell us that the Democrats were the most likely winners all along but it was perhaps only after the further bout of financial turmoil in September that this became very likely.

There are, of course, many reasons why we should not trust the signal from prediction markets completely. The financial stake may not be meaningful (in Iowa prediction markets investments are limited to between $5 to $500 per investor). Relatedly, liquidity may be low and prices thus too sensitive to lumpy trading. And the markets may simply reflect other evidence from opinion polls rather than promote a process of information discovery. But as a device for aggregating information across diverse individuals and for allowing a hedge against unwanted outcomes, these types of markets may well be able to help all markets become more complete. If you happen to be a Republican supporter my commiserations, but if you had also bought those Democratic futures at 53 cents in September you may not have felt quite so miserable this morning!