German Election: Final SMI Forecast

German Election          Merkel to Win, but FDP and AfD will be biggest outpeformers

Using polls as a base scenario, we compare the latest SMI forecasts.  The table below summarizes polls, SMI forecasts and the differentials between polls and SMI forecasts.

Table 1:  German Election:   Social Media Influence (SMI) Forecasts vs Average Polls

Source:  ZettaCap, INSA, Forsa, Forschungsgruppe Wahlen, Allensbach, Infratest dimap, Emnid, Trend Research, YouGov.  Note:  Average of polls taken from polls over previous week.

As previously stated, the large ‘Establishment’ parties are predicted to do quite well in absolute terms.  The CDU and SPD, the two largest parties in Germany and those in the current ruling coalition, are forecast by both polls and SMI to gain in aggregate over 50% of the popular vote.  This will allow them to form a ruling coalition if desired.  However, both parties are forecast by SMI to underperform current polls.  These SMI forecasts are fairly negative considering the forecast for SPD is at the very low end of poll data over the last month whereas the forecast for the CDU is actually lower than any poll figures over the last month.

The AfD’s expected outperformance has come about very late in the campaign.  This is due to the emergence of Alice Weidel as an important leader in the election.  She had been a fairly unknown politician until recently, but has performed extremely well in terms of SMI and has brought her party up in the process.  The big question here is if there will be enough time for improved SMI to work its way into votes as there is unusually a delay.  Weidel is one of the major SMI standouts for the campaign season – not the election, but of the campaign which really only technically includes the six week window before election-day.

The two further left-leaning parties, the Greens and the Left, are expected to disappoint the most.  Both parties were consistent SMI underperformers.  This was somewhat surprising given the topics that dominated the last few years in Germany and globally which included global warming, the environment, and immigration – all of which are normally key topics for these parties.  The inability to generate significant SMI in such an environment is a major red flag.

The real standout of this election is Christian Lindner and his FDP.  We called attention to his increasing SMI early in 2017 and forecasted that Lindner’s strong early traction on social media would result in improved performance for the FDP.  Recall that the party did not surpass the 5% minimum last election and many believed the party would end up withering.  Lindner did the seemingly impossible by resurrecting the FDP.  This process was called very early by SMI ratings.

Currently, SMI forecasts place the final FDP vote at almost 5 percentage points above average polls.  For a relatively small party, this type of outperformance is astounding.  As highlighted in the Disclaimers near the end of this report, there are many reasons why SMI might over-estimate FDP’s actual votes.  One of the main issues is that Germany has an indirect voting system in that the country’s lower house, the Bundestag, eventually elects the Chancellor as compared to other countries where the electorate votes directly for the primary leadership, such as for President in the US or France.  This indirect system will likely diminish the impact of Lindner’s strong SMI as it places higher importance on large well-established parties and career politicians.  In short, a 5 percentage point differential is extremely large, so take it with a grain of salt.  Expect outperformance but understand that Germany’s unique system will likely dampen SMI accuracy to a certain degree.

ZettaCap_German Election_Final Forecast  is the report from where the above excerpt was taken.  This report highlights other alternative data sources such as Google Search Trends, Facebook ‘Likes’ and Wikipedia Page Views as they apply to the German Election.  The other alternative data sources generally confirm SMI forecasts.