SMI: South Korean Presidential Forecast

South Korean Presidential Election          Late Surge for Sim Sang-jung but Moon Jae-in to Win

Using Social Media Influence (SMI) to forecast political races is new and innovative.  It was used successfully to make early forecasts in the 2016 US Presidential Election, where it forecasted an eventual Trump victory, and in the 2017 French Presidential Race, where it forecasted a Macron victory.  At the time of these forecasts, these candidates were still considered long-shots.  As repeatedly mentioned in other posts on this topic, SMI appears to lead other forms of political analysis such as polls, betting markets, and pundits.

As a disclaimer, it should be noted that not only is SMI a relatively new forecasting tool but that previous successes will not necessarily be repeated, especially considering the volatile global environment at present.  Additionally, in some elections, social media data is more limited, as is the case in South Korea where not every candidate is fully active on social media.  With basic disclaimers of potential inability to reproduce previous strong results and limitations on data in South Korea’s case, we can get on with the analysis.

In the 2017 South Korean Presidential Election, there is one candidate who has a SMI rating that far outpaces her other traditional metrics.  In such a case, SMI, as a leading indicator, would ‘pull’ the other metrics higher and we would expect that candidate to surge in the polls and outperform going into voting.  The main limiting factor is time as voting is within weeks so it is unclear to what extent this scenario will play out.

Before getting into forecasts, we should review what has occurred in South Korea over the last few years as it helps to put the election in context.   President Park Geun-hye, South Korea’s first female president, was impeached due to an influence peddling scandal that involved multiple people and entities, but most importantly political and industrial leaders and chaebol, or the country’s most influential companies.  Additionally, Park Geun-hye was no ordinary politician in that she was the daughter of Park Chung-hee, the leader (or dictator depending on who you ask) of South Korea from 1961 to 1979 when he was assassinated.

This backdrop is important for a number of reasons.  First, the current presidential election is coming on the heels of the impeachment.  Second, the influence peddling scandal was very much centered on traditional backroom dealings.  And, third, the impeachment had a strong anti-establishment flare to it.  These points are important in that you would expect the anti-establishment wave, that not only hit South Korea during impeachment proceedings but which also has spread to many countries recently, would impact the current election.

Now, looking at consensus figures for the election:

Table 1:  South Korea Consensus View of Presidential Election based on Polls and Betting Market Data

Source:  Polls (taken as average of polls over last seven days from 12 polls including Hankook Research, MetriX, JoongAng Ilbo, JoWon C&I, Kantar, KSOI, ResearchView, Gallup Korea, EMBRAIN) and Betting Markets (www.predictit.org)

From these consensus figures, it looks like there are really only two serious contenders, Moon Jae-in and Ahn Cheol-soo.  According to polls, these two candidates account for approximately 71% of the intention to vote and according to betting markets nobody else has anything close to a chance.  In fact, the bottom two candidates listed in the table apparently have such low probability of winning that the betting market site used does not even have contracts for them.  Additionally, the election is scheduled for May 9^th, meaning the chances of something drastic occurring before voting starts seems remote.

In this environment, Sim Sang-jung’s SMI has been performing extremely well.  So well in fact that it makes her current poll figures look surprisingly low.  The case of Sim Sang-jung could resemble that of Melenchon in the recent French election.  His SMI was very attractive which was not getting picked up by polls or betting markets at the beginning of the campaign.  Then, his debate performances were greeted enthusiastically and his traditional metrics improved quickly right before voting began.  This late surge for Melenchon, who was a strongly left-leaning politician, impressed and/or scared many observers as he was not considered a serious candidate until the end.

Like Melenchon, Sim Sang-jung has been performing well in the presidential debates which could help to act as a catalyst to improve her standing.  Certainly, her SMI indicates that given the right catalyst she would jump in the polls rather quickly.

As for her policies, they appear to be the least ‘establishment-based’ of any of the main candidates.  In comparison to the other candidates, she has attacked the chaebol more vigorously (which is an advantage in the current environment where they are seen as being involved with the impeachment scandal), has helped to organize labor unions / strikes and worked for labor rights to a much greater extent, is the only candidate in favor of LGBTQ rights, and is the most pro-female candidate by far.  The most important point to recall is that South Korea is going through an anti-establishment period having just impeached an ‘establishment’ president who was caught up in an influence peddling scheme that included other ‘establishment’ individuals including chaebol executives.  It stands to reason that Koreans will look to an anti-establishment candidate, and Sim Sang-jung is the most anti-establishment candidate in the race.

It seems like Koreans are are very interested in what she has to say, given her high SMI, but are not willing to publicly admit support for her.  Averaging at 4.5% in the polls shows a meager support level.  In terms of SMI, however, she is one of the race’s leaders.  The meaning of this high SMI rating in relation to her lower poll figures is that given the right opportunity and sufficient time, she should do well.  She has the opportunity with the ongoing presidential debates.  But, she does not have sufficient time as the election is within two weeks.  As such, the most likely scenario is one similar to Melenchon who produced some ‘unexpected’ waves near the end of the campaign as his polls figures surged but was unable to catch the leaders.

Assuming Sim Sang-jung replicates Melenchon’s path, it still would not be enough as he was able to gain about 5 percentage points in two weeks and about 9 percentage points in the last month of the election.  This would be enough for Sim Sang-jung to gain celebrity, upset many political forecasters and rattle financial markets, but in reality it would not be enough, given her low starting point, to seriously upset the election.

In short, given the unusual gap between a strong SMI and weak poll figures, we should expect her polls to increase into voting.  Further, assuming her SMI can remain strong, we should expect her to outperform expectations on election-day.

As for the rest of the election, it is expected to play out more or less according to consensus.  SMI ratings for Moon Jae-in and Ahn Cheol-soo place them in first and second place respectively, just as poll and betting markets do.  Put another way, if Sim Sang-jung was not running, SMI would roughly agree with polls and betting markets and the election would likely run its course in-line with consensus.

There is a chance that Social Desirability Bias could be at play in South Korea.  This was the case in the US as many did not want to publicly admit supporting Trump which tended to distort polls.  In France, Social Desirability Bias appears to have had a much smaller impact on polls, so it depends on the election.

For Social Desirability Bias to take hold, there needs to be the perception of some candidates being more socially acceptable than others and for people to feel enough social pressure that they will modify their true answers of who they support.  In South Korea, this could be the case as Sim Sang-jung does not fit the traditional mold of a politician and, further, South Korea is a fairly homogeneous and conservative political and economic culture.  In fact, many of her policy stances are progressive in the Western model of politics, which likely would not be seen as socially acceptable in a traditional political environment.  Supporting LGBTQ rights, attacking chaebol, helping to organize labor strikes, and highlighting the dangers of a large wealth gap in society are not topics that many in a conservative political environment would see as socially acceptable.

The facts that her positions and demeanor are likely seen as less socially acceptable, that she is making a significant impact on social media, and that she is not doing well in polls, provides enough evidence showing that Social Desirability Bias is possible.  Assuming this is the case, her outperformance on election-day could well exceed what occurred in France for Melenchon who outperformed his polls but by less than a few percentage points.

As for now, the best way forward is to continue to monitor the SMI going into voting.  The base-case scenario is that Sim Sang-jung will be the candidate who will outperform to the greatest extent but that Moon Jae-in will end up winning the election.  As for financial markets, minor turbulence are forecasted as the most left-leaning candidate makes some last minute waves, but expect things to end up with the consensus candidate being elected.