Brexit has received a lot of attention during the U.S. election campaign and it's now having an impact on how investors and pollsters approach next week's vote.
Britain's surprising decision on June 23 to leave the European Union caught the world off guard because last-minute polls and betting markets put the Remain camp ahead. The stunning result sent global markets reeling and the value of sterling down by more than 10 per cent.
While the U.S. election is a different kind of campaign, there are enough similarities to make investors and analysts far more cautious.
Like the Remain side in the Brexit campaign, Hillary Clinton has been consistently ahead in most polls for weeks and betting markets give her about an 80 per cent chance of winning. And like the Vote Leave side in Brexit, pollsters are having a harder time measuring Donald Trump's support.
And all that means no one is taking any chances this time.
"I do think some lessons have been learned from Brexit, undoubtedly," said Craig Erlam, a London-based analyst at currency trader Oanda. "When it came to the days leading up to Brexit, the way markets were trading, you'd think that the votes had been cast, you'd think the results had been read out and the markets were 90 per cent of the way to being priced in. That suggested people saw the Brexit vote as a foregone conclusion."
Mr. Erlam said he's seeing more caution when it comes to the U.S. election. For example, trading in the Mexican peso, which is believed to be more vulnerable to a Trump victory, has moved sharply with opinion polls, and there are reports that foreign fund managers have been pulling money out of the U.S. markets despite Ms. Clinton's lead, which is supposed to be good for investors. "This suggests to me that the markets do not see this in the slightest bit as a foregone conclusion," he added.
Pollsters, too, have been debating the fallout of the Brexit vote for months, and many have been rethinking how they track support. John Curtice, a professor of politics at Strathclyde University in Glasgow and a leading researcher on polling, said American and British pollsters have been grappling with sample sizes.
Prof. Curtice explained that pollsters have to make assumptions about the number of people they poll from different demographic groups such as gender, race and age. During the Brexit referendum, education became a much more important factor for measuring the support for Vote Leave and Remain (in general, people with less education supported Leave and those with higher education, Remain). If the sample size wasn't right, polls could be misleading.
"Public opinion polls in the U.K. do not routinely collect educational attainment. Therefore there was quite a lot of scrambling before June 23, as opinion polls increasingly began to ask people questions about educational attainment in order to make sure they didn't have too many, or too few, graduates," he said.
Pollsters in the United States have had to adjust as well to cope with many of the same challenges tracking Trump supporters, he added. "I understand that some pollsters in the States have been worrying about 'Do we have the right number of graduates and high-school leavers in our samples or not?' … Insofar as Trump is getting a disaffected white, blue-collar, low-educational-attainment working-class vote, that is similar to Leave."
There is also a continuing debate over the difference between telephone and online polls. During the Brexit referendum, phone polls tended to show more support for Remain while Internet polls had Leave ahead. Prof. Curtice said it still isn't clear why there was such a difference.
Judah Grunstein, editor-in-chief of World Politics Review, which follows global political trends, believes the polls in the United States will be more accurate than those done during the Brexit vote in Britain, as well as last month's referendum in Colombia, where polls showed strong support for a peace deal with FARC but the final result went against the deal.
Referendums are one-offs, he said, whereas the U.S. vote has historical patterns that can be measured. The U.S. electoral system also puts more emphasis on state-by-state results, meaning national polls are less important. And referendums can be complicated by voters' attitudes toward the government in general, which can take precedence over the issue at hand.
"For all these reasons, I'd argue that the polls in this case are more dependable than the ones in the run-up to Brexit," he said. "But of course, we'll only know for sure when the results are tabulated next Tuesday."