By Justin Wolfers, nytimes.com, Oct 11, 2014
How would commentary on the midterm election look if economists, rather than Beltway pundits, were calling the race? You would read a lot less about personalities, gaffes and gossip, and a lot more about fundamentals like the state of the economy. And you would certainly get a more sophisticated reading of polls and political prediction markets.
Campaign events often drive sharp shifts in sentiment, although these effects are typically short-lived. A result is that polls often fluctuate wildly even as electoral fortunes evolve slowly. But as Election Day approaches, polls tend to shift toward the fundamentals, as economic and political realities come into sharper focus. There’s no point being steered in the wrong direction by polls that will later reverse themselves, which is why careful poll analysis filters polling data through a lens that takes account of the underlying fundamentals.
How should you combine insights from the basic political factors — the state of the economy, opinion polls and voter expectations? Economics doesn’t provide a formula, but rather a suggestion: Turn to political prediction markets where people trade securities linked to election outcomes. When many people make trades based on their individual perspectives, the price comes to reflect each of the most relevant factors, effectively weighting each by the confidence that traders have in them.
Even though markets are prone to their own failures, they have amassed a better record of accuracy than even the most sophisticated models that are based on fundamentals and polling. The point is that while markets aren’t perfect, in practice they’re less imperfect than the other election forecasters.