French Election                  First Round of Voting

The 2017 French Presidential Election has been ‘called’ by consensus commentators and betting markets many times already.  For much of 2016, most commentators assumed that the Republican candidate (eventually Fillon) would win the election.  Then, as France and Europe experienced issues with refugees and/or recent immigrants, the media and betting markets favored Le Pen and her populist party especially in the first round of voting.  Macron, an Independent promising a new start for France, pushed both aside in early 2017 to become the next ‘obvious’ front-runner.  Now, in a very late surge, far-left Melenchon has thrown a wrench into forecasting this election and commentators talk of being ‘dumbfounded’ by a tight race with various unexpected names in the top cluster of four candidates.

Social Media Influence (SMI) ratings showed a very different story than that portrayed by the media and betting markets.  Roughly summarizing the SMI interpretation of the election by main candidate:

• Fillon – was not seen as a true front-runner by SMI, even when polls and betting markets showed him as such. He was quickly identified as a relatively weak candidate from early December, which was well before any scandals were revealed.  Those analyzing the election normally mention the alleged family-payment scheme as the reason for his decline, but SMI showed that, even while his poll and betting market figures were very high, he lacked real support.  In short, SMI correctly disregarded him early-on and SMI does not forecast him outperforming polls going into the first round.
• Le Pen – did not produce outsized influence on social media and was identified early in December as likely to decline. Even with substantial and negative Europe-wide headlines concerning refugees and/or recent immigrants during the last leg of the campaign, this populist anti-immigrant candidate was not able to fully capitalize on such events in social media which made her candidacy look increasingly shaky.  Going into voting, her SMI has been on a rather steady decline over the last month and she is not expected to outperform polls.
• Macron – had an impressive SMI during November, December, January, February, and March. We called in early December that his poll figures would surge and that he would win the election.  Eventually, polls and betting markets caught up with SMI in declaring him the front-runner and his position looked very strong going into the debates.  Post-debates, however, his SMI has declined putting our December forecast of a Macron victory in doubt (see later in this post for updated forecasts).
• Melenchon – was identified in early December as being the ‘dark horse candidate’ as his SMI was extremely strong in comparison to his meager poll and betting market figures at the time. He was the candidate who, given the right opportunity, could quickly gain traction in polls as his SMI was surprisingly strong.  His SMI surged during and post-debates and going into the first round of voting his SMI is the strongest of any candidate, having just recently surpassed Macron.

In general, SMI has done an exceptional job during the 2017 French Presidential campaign.  It identified, before any other known forecaster, tool or market, that:

• Neither of the traditional / main political parties (Republicans and Socialists) would field strong candidates,
• Le Pen would likely disappoint and end up declining during the campaign,
• Macron would surge to become the front-runner, and
• Melenchon could make it to the second round, surprising most.

The main sticking point at present is the fact that Melenchon’s SMI has actually taken the lead right before the first round of voting.  To be clear, SMI identified Melenchon as being a substantial and unrecognized threat to the group of leading candidates from early December.  So, his surge was not a surprise to those following SMI.  However, the strength of the surge and the general volatility that has ensued was underappreciated.  There are many variables to take into account outside of absolute SMI levels which provide insights into the best way to interpret late SMI movements, as explained next.

Undecideds can play an important part in determining the election

As with many other elections having taken place during the last 12 months in different countries, there appears to be a tendency for French voters to vote outside traditional party lines.  A partial result is that many voters remain undecided until the very end, perhaps hoping to get some new information or perhaps simply not wanting to declare their intention as they intend to vote for someone not considered socially acceptable within their social circle.

According to press reports, undecided voters amount to approximately 33% to 40% of the voters.  This is extremely high by any standard.

If correct, this election is still very much in play.  According the most recent polls, only 5 percentage points separate the poll leader (Macron) to fourth place (Melenchon).  With about a third of the electorate undecided, 5 percentage points could swing very rapidly.  Put another way, the late deciders could determine the election – that is if their distribution is not similar to the current polls.

SMI shows underlying influence of the candidate in social media and is used to determine support for the candidate in an unfiltered and unbiased way.  In the French election, the main candidates’ SMIs vary to a significant extent from their poll figures, especially during the last month when SMIs have been more volatile than expected.  A preferred interpretation of the divergence between polls and SMI in this election is that SMI being stronger than poll results indicate a candidate likely to attract a larger than expected proportion of undecideds.  Conversely, an SMI showing a candidate in a worse position than poll results indicates that the candidate will likely underperform in attracting late deciders.   Again, in this election, this could make enough of a difference to upset the race.  In short, late surges in SMI should be viewed as an important indicator for the determination of voting intentions of late deciders.

How significant are late surges in SMI?

It seems that a purely event-driven surge should be somewhat discounted.  For instance, if a surge occurs around a specific event, you can more or less assume that once that event fades so too will the surge.  We saw Le Pen’s SMI increase substantially around major negative headlines concerning recent immigrants, but once these events ceased making headlines her SMI began to fall back to its pre-event zone.

Melenchon’s SMI began to really surge during the first French Presidential Debate.  This could be seen as an event-driven response, but his SMI gained further strength after the second debate.  Presently, his increase appears much less event-driven as his SMI has gained further after the debates as other candidates, who had also benefited from the debates, sank.  In fact, Melenchon is ‘sucking the oxygen out of the room’ as other candidates are having trouble making an impact as Melenchon has become so influential, at least in the very short-term.

If his SMI slipped right before the election, it would signal that perhaps his recent popularity was based on fickle voter interest or novelty.  However, it is near, though not at, his 2017 high on the eve of voting.  In contrast, Macron’s SMI is currently well below his January to April 2017 average.  The SMIs of Le Pen and Fillon are around their respective mid-ranges over the same period.  In other words, social media is pointing to Melenchon gaining significant traction as Macron is struggling right at the end of campaigning. (This should be put in context as Macron is ‘struggling’ in the sense that his SMI has come off significantly from its high but is still solidly in second place.)

Melenchon’s late surge does not appear to be purely event-driven and his SMI has remained high after the debates have well passed.  This surge is to be taken seriously, but perhaps not at face value.

As Melenchon’s surge is so late, can it be taken at face value?

In the US election, SMI ratings were taken at face value but in that race the candidates and parties had fairly well established trends.  Especially as the general election got close, the volatility of the SMIs decreased.  In France, SMI volatility has increased in the last month before voting starts which makes the stickiness of Melenchon’s SMI surge more questionable for the general election as it has not had sufficient time to prove itself as stable at higher levels.

Certainly, Melenchon’s current high SMI should be interpreted to mean that an unusually large proportion of the late deciders and swing voters will tend to vote for him.  Put another way, as his current SMI is so much higher than his current poll figures, we should assume that he will be able to attract a surprisingly high number of undecideds.  The result will be general outperformance of poll figures.

Taking Melenchon’s SMI at face value (meaning his current superior SMI will translate into the highest level of votes in the first round) is questionable under current circumstances.  Though less technical of a response then you might be looking for, it seems that (given the volatility of the SMI late in the race, the fact that undecideds are rather high, and his relative lack of an established party-base) Melenchon’s SMI will mostly impact late deciders.  In other words, taking SMI at face value is relatively risky here.

Le Pen to Underperform

The stars aligned during this election for Le Pen and her SMI still did not manage to take the lead at any point from November 2016 to April 2017.  Among her many advantages going into 2017:

• She had by far the largest social media following,
• Her domestic and international media attention was phenomenal,
• Her domestic and international name recognition exceeded all other candidates,
• Populist upsets in the UK and US set the stage for a repeat in France,
• Anti-establishment political trends, which seemingly would help her, were strong globally,
• Current events in France and Europe portrayed recent immigrants and/or refugees in a negative light (supporting Le Pen’s view),
• The two main parties in France, Republicans and Socialists, were more or less knocked out early in the race due to scandals and ongoing unpopularity of the sitting (Socialist) president, and
• Her party’s infrastructure was superior to all other non-establishment parties.

Given the unusual coincidence of such supportive factors occurring around the same time, a rational individual would assume that it was Le Pen’s race to lose.  And, many political analysts and commentators did just that.  But, her SMI struggled throughout the campaign, moving higher during periods of recent immigrant upset but then slowly sliding back down again.

In other words, if Le Pen was unable to excel in these circumstances it does not seem likely she will receive an ‘unexpected’ boost in the voting booth.

Late Shooting

ISIS has claimed responsibility for a deadly attack on a police bus in Paris just days before the first round of voting.  The knee-jerk reaction would presumably be in favor of Le Pen whose party supports greater security measures.  And, her SMI has moved higher but not to a significant degree.  In fact, Melenchon’s SMI surge due to his debate performance dwarfs likely terror attack based bumps for Le Pen.  In short, this late shooting should help Le Pen but less than most believe.

Revised Forecasts

SMI continues to forecast Macron gaining the most votes in the first round of voting.  As for second place, it is too close to call.  Melenchon, with his strong SMI, will disproportionately benefit from gaining late deciders which will pull his already upward trending poll figures higher into voting.  Le Pen’s loyal base will vote but her SMI has depicted a relatively low level of influence, especially given current events, on social media implying that late deciders will not likely break for her.  Fillon’s SMI has been consistently low throughout the campaign and does not indicate he will be able to outperform his current polls and is not seen as a real contender going into voting.

Essentially, we have Melenchon surging in SMI and polls right at the end of campaigning.  In contrast, Le Pen’s SMI has trended mostly below her polls throughout the campaign which has helped to pull her poll figures slowly lower.  The real question, which neither SMI nor polls can correctly predict at present, is how much of Melenchon’s recent momentum will carry over into voting.  Assuming that the late volatility in SMI will directly influence undecideds, results in a dead heat between Melenchon and Le Pen.

Most Likely Scenarios for Second Round

The most likely scenario according to betting markets and polls is Le Pen versus Macron in the second round, with Le Pen having gained the most votes in the first round of voting but losing to Macron in the final.

Surprisingly, betting markets are only seriously considering this scenario.  There are individual contracts for Le Pen and Macron making it as individuals to the second round of voting – and each of these contracts place the probability at over 80% (see table below).

According to SMI, there are two likely scenarios.  The first is very similar to the consensus view described above with the exception that SMI forecasts Macron, and not Le Pen, gaining the most votes in the first round.  In the run-off second round, Macron will beat Le Pen.

In a second likely SMI-based scenario, Melenchon edges out Le Pen and moves to the second round versus Macron.  In this case, we would need to come back and see how SMI trends develop during second round campaigning.  As often mentioned in this post, Melenchon’s SMI has surged recently and its development, post first round, would indicate voting indications for the second round.  This scenario is much less clear and much more risky than the first scenario.  Recall as well that Melenchon did very well in the debates, which worked as catalysts for his SMI, and Macron’s SMI began to stall around the same time.  The fact that there is an additional presidential debate before the second round of voting bodes well for Melenchon who appears to directly benefit from this platform.  Assuming a Melenchon versus Macron second round, SMI ratings should help to provide insights to cut through the froth.

Potential Surprises

The three main potential surprises of the first round are the underperformance of Le Pen and Fillon and the outperformance of Melenchon.  Before going into specifics, we should first check the betting or prediction markets to establish the current consensus.

Table 1:  Betting Market Based Probability of French Candidates Succeeding in the 2017 French Presidential Election

Source: PredictIt

Betting markets and polls have shown Le Pen in a very strong position throughout the race.  Generally, she appears in betting markets and polls in first or a strong second place showing.  However, as previously discussed, Le Pen has benefited from an extraordinarily positive environment and her SMI was never able to take the lead at any point in 2017.  This negative divergence shows that her underlying support, especially outside her core base, is not that strong.  Having fallen into third place in SMI, the proportion of undecideds voting for her is expected to be rather small in comparison to her relatively high first round poll figures (which show her alternating the lead with Macron).  The expected lower percentage of undecideds voting for her will push her performance slightly under expectations.

Fillon appears to have clawed his way back to a solid third place according to the betting markets.  SMI shows that his position is actually fairly weak as his SMI has been considerably lower than his poll standings during the campaign.  Outside of his loyal following and his solid party base, he does not appear to have generated any significant enthusiasm on social media.  Expect Fillon’s vote tally to suffer and for him to generally underperform current polls.

In contrast, Melenchon, though having received a considerable amount of sensational media attention is, according to betting markets and polls, not seen as a real contender as the probability of him becoming the president is only at 8%.  Melenchon still has a lot of room to outperform expectations.  As his SMI is currently the highest going into first round voting while most polls show him in fourth place.  This positive divergence (as opposed to Le Pen’s negative divergence) bodes well for him to outperform expectations.

Disclaimer

Recall that the SMI and most every tool or technique that does not have considerable backtests should be considered experimental.  SMI forecasts in the 2016 US Presidential Election were extremely good, being the only forecasting method to correctly call Trump winning with 306 electoral votes.  That being said, take care with this and any forecasts.

Additionally, SMI ratings during the 2017 French Presidential Election have been rather volatile going into voting.  As previously mentioned, late volatility makes SMI forecasts less accurate.  The best case is when trends are fairly consistent.  They do not need to agree with polls, betting markets, or political analysts, but it does help when SMIs trend and do not move quickly in either direction, especially so close to voting.  In France’s case, the SMIs of the candidates shifted very quickly within the last month of the campaign before the first round of voting.  So, again take care when using these forecasts