Polls are still reliable, and they show Marine Le Pen losing

France’s chances
by Natalie de Vallieres and Biodun Iginla, Political News Analysts, The Economist Intelligence Unit, Paris

The nationalist leader is still unlikely to become president of France

Europe

Feb 22nd 2017

JOURNALISTS often joke that three examples make a trend. Following the votes for Brexit and Donald Trump, a victory by Marine Le Pen of the National Front (FN) in France’s presidential election would complete the anti-globalisation trifecta. She has dominated the polls ever since news broke that François Fillon, her centre-right rival, had paid his wife and children about €1m ($1.05m) over the years for jobs critics call fake. But a deeper analysis shows that Ms Le Pen is more likely to end the streak than to continue it.
After last year’s surprises, many people stopped trusting polls. This is misguided: in both cases, surveys correctly predicted that the race would be tight. If polls in France are similarly reliable, Ms Le Pen’s chances in the first round of the election are excellent. The Economist has aggregated 100 French polls (a technique that is still rare in France, though it is de rigueur in Britain and America). We find that if the first round were held today, Ms Le Pen would carry 26.1% of the vote. Emmanuel Macron and Mr Fillon would trail with 19.7% apiece.
These figures could change, but big shifts are rare. According to a database of French polls since 1965 compiled by two political scientists, Will Jennings and Christopher Wlezien, surveys 60 days before the first round have been off by just three percentage points on average. Using this record to run 10,000 computer simulations shows Ms Le Pen as the heavy favourite. She wins the first round 77% of the time, and is a 96% shoo-in to reach the run-off.

The race for second place is much tighter (see chart). Mr Fillon’s chances of reaching the run-off have fallen from 79% to 50%, slightly more than Mr Macron’s 47%. Benoît Hamon, the Socialist candidate, manages just 5%.
However, the second round is a different story entirely. When voters are asked to pick between Ms Le Pen and Mr Fillon, she loses by 13 percentage points. Against Mr Macron, it is 20. At this stage, voters tend not to change their minds: in presidential elections since 1981, the average poll of a potential run-off 70 days out has also missed by only three points. If they are similarly reliable this time, Ms Le Pen has less than a 5% chance of victory.
Of course, unusual events cannot be ruled out, and many voters are still uncertain. Betting markets give Ms Le Pen odds of 28%-43%. Punters may think further scandals could fell whoever faces her in the second round. Should it be Mr Fillon, leftist voters who dislike him might stay home. But such a drop in participation would have to be huge to matter. If the polls hold, even if every FN supporter actually votes, a fifth of opposing voters would have to drop out for Ms Le Pen to win. That is vastly larger than the shifts in Britain and America.
The most likely outcome is that history will repeat itself. In 2002 Jean-Marie Le Pen, Marine’s father and the FN’s founder, snuck into the presidential run-off, only to lose by 64 points. Just 14 months ago, the FN topped national first-round polls for regional elections. But its opponents teamed up, and it failed to win a single region. Perhaps this time will be different. But if Ms Le Pen wins, it will be a far bigger shocker even than the votes for Brexit and Mr Trump.