I’ve just returned from the European Political Science Association conference in Barcelona. One of the more interesting papers I saw was a study of the effect of polling on campaigns. In the paper, Philipp Denter and Dana Sisak note that dozens of democracies have bans on polling late in the campaign season. In France, for example, no poll may be published within seven days of an election. Switzerland has a ten-day embargo. Up until 1996, Italy banned polls for a month prior to elections. (See this report for more details.)
This variation allows us to examine the effect of polling on elections. As Denter and Sisak suggest, polling, by removing some uncertainty about a race, tends to firm up the race in favor of the frontrunner and generally results in a reduction of campaign expenditures and competitiveness. From their abstract:
Under a simple majority rule, there always exist equilibria in which the initially more popular candidate invests more in the campaign and thereby increases his lead in expectation. In other words, polls create momentum for the favorite. Only when campaigning is very effective and the race is very close, a second type of equilibrium may exist: in these situations the trailing candidate also has an incentive to out-spend and overtake the leading candidate. Regardless of the type of equilibrium, polls have a tendency to decrease expected total campaigning expenditures by amplifying ex-ante asymmetries between candidates and thus defusing competition.
[Cross-posted at Mischiefs of Faction]