There haven't been any new GB-wide EU referendum polls published since Wednesday, but Friday did see a rather widely reported intervention from the President of the United States… There followed the inevitable suggestions from some that this was game over. It’s not game over, and it won’t be until the votes are counted just under two months from now, but it certainly seemed like a significant moment. What effect might it have?
Earlier in the week I had been reminding a group of investors that risks to the central forecast ran in both directions. It’s possible that an event like a terrorist attack, an escalation of the refugee crisis or a sovereign debt flareup occurs to help Leave. On the other side, establishment use of heavy artillery was always likely and is essentially factored in already. But the Leave campaign’s claims about suggestions about US trade policy being dismissed by that country’s President wasn’t, so the next set of polls will be interesting. Very little polling has been done since the Treasury dossier, let alone the Obama speech, so you can bet that someone will have been in the field over the weekend.
So far polls have tended to focus on whether the US president was right or wrong to intervene, plus some “more or less likely” type questions (see ComRes and YouGov). Whether you think he should or should not have given his opinion doesn’t necessarily tally with will or will not be influenced by it. And “more or less likely” questions, even if respondents were to provide an accurate measure of their probability of voting a certain way (a big if) still take no account of how likely/unlikely they were to do that thing in the first place.
Basically there are two ways that Obama’s comments could influence polls, the first of which, if it happens, will do so immediately. There could be a direct impact on the polls, partly from the impact on wavering voters, but also perhaps from the classic event-driven overshoot.
But the other avenue is the “bigger picture” effect. Few if any interventions will be game changing on their own. But the drip, drip of warnings will collectively take their toll (HT Janan Ganesh for articulating this). “Project fear” is a battle of attrition, not a single bazooka shot.
The major caveat to this is that most political events have less of an impact than Westminster expects them to (see Panama). But this time impact on the betting markets has been substantial, with Odds of Brexit lengthening to imply a probability of just 25%, compared with 37% earlier last week:
Can't hear Obama speech from here but whoever spent £2.5M bidding up the #Brexit on Betfair this wk just got hit with a baseball/cricket bat
— Matt Singh (@MattSingh_) April 22, 2016
In fact a serious journalist actually asked me over the weekend the probability of Remain’s vote share exceeding the 67.2% that Harold Wilson led the Yes side to in 1975. That’s tricky to estimate, because it’s very unlikely, and we have less experience of unlikely events, because by definition they happen less often than likelier outcomes. Or to be technical about it, we can’t be as sure about the precise shape of the probability distribution in the tails, because we have considerably fewer data points there.
But the fact that the question is even being asked is itself telling. For what it’s worth, the model suggests – at face value – about a 10% chance of such an outcome, the betting markets about half that.