THE US midterm elections are just days away, with as yet no clear indicator as to the fate of the Republican-controlled Congress. While there is a consensus that Democrat control of the House of Representatives is very probable, there is by no means unanimity. As for control of the Senate, it looks likely to go down to the wire. Wouldn't it be nice if, days or weeks in advance
THE US midterm elections are just days away, with as yet no clear indicator as to the fate of the Republican-controlled Congress. While there is a consensus that Democrat control of the House of Representatives is very probable, there is by no means unanimity. As for control of the Senate, it looks likely to go down to the wire. Wouldn't it be nice if, days or weeks in advance, we were able to predict with accuracy the outcome of elections? Political polling results are notoriously fragile. Traditional methodologies struggle to keep up with demographic shifts, technical trends, political fragmentation and polarisation. Moreover, polls not only have a significant (3-4 per cent) margin of error - the margin itself is subject to its own dynamics and doesn't necessarily give us a good indication of the range of possible outcomes.
So where do we go? One thing we know: in general, markets do a pretty good job of allocating and informing investors. Despite what fund managers and investment gurus say (and charge), it is hard to outperform the market - the best prediction of the level of a financial asset tomorrow is its value today. Keynes had it right when he noted that the stock market was in some ways akin to a beauty contest; the object is not to pick the face one finds the most beautiful, but the one you think other people will, on average, find most beautiful.
So what? Well, there are financial markets, real money markets, which deal in these future events. For one, in particular, its accuracy is startling. The Iowa Electronic market is a real money market, where investors get a real money reward for their predictions. Investors trade in vote shares and political events in a number of elections, not just US elections. There are good reasons why such a market would be expected to have at least a decent chance of giving as good a prediction as a poll. Markets exist not just to allocate resources, they also exist to provide a mechanism for the aggregation of information. Markets operate efficiently in terms of this information aggregation where a number of events combine, as they do in the case of the Iowa political market.
Thus, there is a clear, definite action to be predicted (the outcome of an election); there is a real monetary reward for being accurate and cost for being inaccurate (the payoff is not in play money but in real money); there is anonymity in trading; there is liquidity (the ability to sell and buy) from the same volume of trade and there is a fair trading mechanism.
Technically, legally and in regulatory terms, the Iowa market is a futures market. Futures markets are markets where investors trade contracts for goods, such as pork bellies, oil, currencies etc, with the delivery (or more generally a cash settlement) taking place in the future.
Typically futures markets are called month-
commodity, such as March-Wheat, which means that trading now involves predicting what the price will be for wheat at the end of March.
Modern finance theory has achieved a great deal of understanding about how and why futures markets prices evolve. Remarkable as it seems, the price of a commodity on a futures market is, in cold hard statistical terms, the best predictor of the actual cash price that will be paid for the commodity in the future. Thus a futures contract on June IBM shares trading at $120 implies that the market expects that at the end of June, IBM shares will be $120. There are technical proofs of this issue, but the essence is this: the futures price is the best predictor of what the actual price will be in the future. In the Iowa market, traders trade on contracts which pay off on the relative vote share of individuals or parties. Contracts pay an amount equal to the fraction of the popular vote received by a candidate times $1.
By contrast to polling, a futures market explicitly forces traders to consider the outcomes in a dispassionate and calculating manner. Thus, the prediction of the market, especially as the date of the election looms, is substantially closer to the actual outcome than opinion polls and generally lies within 1-2 per cent of the actual outcome. We should not be surprised, especially because it is a well-known feature of futures prices that they converge to the actual price of the commodity in question.
However, the remarkable thing about the Iowa Electronic Market is that it appears to have good predictive power not just immediately prior to the poll but months prior.
While eve-of-poll predictive power of voter markets is known to be high, recent research has indicated that the predictive power of these markets is greater than polls even up to three months in advance. From August 2004 and from early October 2000, traders on the IEM were consistently predicting a Bush victory, in the face of polling numbers that suggested that Kerry and Bush were tied and in 2000 that Gore led by 6 points.
So what of the midterms? Since September, the data have consistently suggested that the Republicans will lose the House but retain control of the Senate. One problem with the markets, however, is that while there is in effect a national election for the House, the Senate election is smaller and more fragmented. The evidence suggests that the larger the electoral pool the greater the accuracy of the market. That said, the smart money suggests a split result: House Democrat, and Senate Republican. The Iowa Electronic Market is at www.biz.uiowa.edu/iem/
Brian M Lucey is a senior lecturer in finance at the School of Business, TCD