The last two weeks have seen an apparent widening in poll results in favour of a Remain win, with polls more often showing a 10-point lead than not.
This has prompted a shift in sentiment towards the British pound which has rallied accordingly.
Commonly-reported moving averages of opinion polls show this pick-up in Remains chances:
Above: Balance of 'Remain' on the 6 month poll moving average versus share of telephone polls.
This does not reflect a shift in view of the electorate, but rather a shift in polling methods.
Looking through some of this noise, a more stable pattern emerges that suggests little change in view by voters.
Going forward, we recommend tracking developments in online and telephone polls separately.
A moving average of recent polls for the June 23rd vote on the UK’s membership of the EU shows a big move in favour of a Remain win, with the media now frequently reporting a Remain margin of about 10 percentage points.
However, it is likely that this does not reflect a shift in the electorate, but rather a shift toward more frequent telephone polling, which tends to be biased towards a Remain win.
There is an Observable Bias Between Telephone and Online Polls
There is however a distinct wedge between telephone and online polls.
In fact, the average of the most recent six polls average shows a 10-point wedge, with the Remain side being favoured in telephone polls by 9 points, while the Leave side is favoured to win in online polls by 1 point.
There are many possible reasons for this divergence, including the fact that online poll respondents tend to be self-selective, with the polls distributed more easily via social media (something the Leave campaign is using more actively owing to a smaller budget), while telephone polls tend to have a more captive audience selected by the polling companies to accurately represent the population.
This bias is confirmed by two recent sets of polls, conducted in late February and mid-April.
In both instances, the same polling company conducted the same poll separately with an online audience and a telephone audience.
In both cases, the Remain margin was wider in the telephone poll—by a margin of 17 points in February and 8 points in April.
Given the bad rap that polls have had in recent years (especially in the UK, where they failed to predict the Conservative majority election win in 2015), it’s not clear who to believe.
But we tend to favour telephone polling methods slightly over online polls, in part because of the selection bias that online polls can attract.
But there are valid criticisms of telephone polling too, and the truth likely lies between the two methods.
Recent “Leave” Poll Wins are all the Same Polling Company
Even within the polls themselves, there are interesting patterns. The weekly polls conducted by ICM are the only polls conducted since April 1st to favour the Leave side.
All other polls have pointed to ties or Remain wins.
Where to from Here?
Betting odds in favour of a Remain win have increased sharply in the past ten days.
Anecdotal evidence suggests that President Obama’s Remain stance led to those in betting markets putting a higher probability on a Remain win.
But we do wonder if the apparent surge in opinion polls is being (incorrectly) reflected in betting odds as bookies think that support for Remain is increasing sharply.
Despite the rise in odds, we emphasise that we remain cautious going forward—there are still nearly two full months between now and the referendum.
As our recent analysis has shown, while a Remain win is favoured by both polls and betting markets, micro-analysis of polling data suggests that turn-out will be key, and a low–to-average turnout will favour a Leave win.
The rise in betting odds led us to take profit on our Dec16/Dec17 short sterling steepener trade today, but on the FX side we maintain our view that GBPUSD will continue to depreciate into the vote date, with a target of 1.35.
Going forward, we’re likely to see more noise in the polls. As polling companies appear to be shifting towards more frequent telephone polling (at least in recent weeks), expect to see sizeable Remain margins more frequently in the poll reporting, and “polls of polls” to consequently reflect this. Betting odds are likely to come off their recent highs, too.
Ultimately, we advise caution in reading too much into this recent drift in polls, and recommend looking at developments in online and telephone polls separately in assessing what will likely be a very close vote.