Blog

US Elections: Why a Replay of 2016 Is Less Likely

With less than forty days to go to the US presidential election, latest polling averages continue to favour a Biden victory with the former Vice President enjoying a comfortable 5-point lead over President Trump. However, haunted by the ghosts of 2016, the general public seems to be much more critical of public poll forecasts. In an August Pew Research Center Poll, Biden’s support rate was 8 points higher than Trump, yet more surveyees believed that Trump would win the election. In another Fox News Poll, a larger portion of voters thought more of their neighbours would vote for Trump, despite a stronger support rate for Biden.

Could polls be wrong again such that Trump stages another election night surprise? By analysing state-level polls, news sentiment trends and voter migration patterns, our short answer for now is “No” – the 2020 election is unlikely to be a replay of 2016.

Our by-State Monte Carlo Simulation Model Suggests an 82% Probability of Biden Winning

Where many polls got it wrong in 2016 was the voting outcome in several swing states, especially in three historically Democrat-held states – Michigan, Pennsylvania and Wisconsin – where Trump narrowly beat Clinton, securing 46 decisive electoral votes.

To better understand the election odds this time round, we ran a Monte Carlo simulation model using state-level polls, which we adjusted for historical partisan lean and mail-in voting rules (we assumed that more difficult mail-in rules would benefit Trump, as they could discourage turnout for Biden voters who are more likely to vote by mail). Currently our model points to an 82% probability for a Biden victory, as shown in the chart below.   

Supporting our model forecasts are Biden’s stronger and more consistent lead in the three states that flipped in 2016, as well as better polling in other battleground states like Arizona, Florida and North Carolina. In other words, while state-level polls could be off again, they need to be wrong by a much bigger margin than 2016 for Trump to win the national election.

Biden’s Polling Lead Has Been Less Sensitive to Sentiment Swings

Unlike Clinton, Biden likely also benefits from his lack of major controversies. By looking at relative news sentiment of the candidates compiled from over 80,000 news articles, we found that Biden has consistently enjoyed more positive sentiment than Trump. In contrast, Clinton was faced with a prolonged period of more negative media coverage in H1 2016, which likely cost her dearly in terms of poll ratings.

In addition, Biden’s polling lead has been much less sensitive to the fluctuations in relative news sentiment, which could suggest that most voters have already made up their mind and are less likely to be swayed by short-term news cycles. The last point is also likely a reflection of increasing polarisation of the electorate and bipartisan politics. As shown below, our topic modelling analysis of speaker transcripts at the Democratic National Convention (DNC) and Republican National Convention (RNC) reveals very different dominant themes and campaign ideologies.

Voter Migration Patterns Do Not Support the Theory of “Undercover” Trump Voters

One popular argument about why Trump could score another surprise victory is that there could be a hidden bloc of Trump supporters who are hiding their true voting intentions given social pressures. Could this theory be true?

We took a deep dive into data from the Democracy Fund + UCLA Nationscape surveys, examining detailed questionnaire answers of over 262,500 registered voters collected from weekly surveys between July 2019 and July 2020. From the analysis we found that over 9% of voters who self-declared to be Trump voters in 2016 indicated that they would vote for Biden this time round. In contrast, less than 4% of 2016 Clinton voters suggested that they would switch to Trump in 2020. The gap between the two measures has been consistently wide over the whole study period. In our view, this finding goes against the theory of hidden Trump voters, as people who already indicated their support for Trump in 2016 would have less incentive to cover up such support in the same survey.

Looking further into survey responses from this group of Trump-Biden switch voters, it seems disappointment about Trump’s performance has been a major factor driving the switch – majority of them disapproved of the way Trump handled his job as president, with the proportion of “Strongly Disapprove” increasing substantially since March 2020, likely following the Covid outbreak in the US.

Fewer Third-Party Voters Might Work Against Trump

Third-party candidates received over 5% of total voters in the 2016 election, with Libertarian candidate Gary Johnson and Green Party candidate Jill Stein polling as high as 10% and 5% respectively in October. Two of them collectively won 3-4% of votes in Michigan, Wisconsin and Pennsylvania, where Clinton lost by less than a point and the states flipped. However, the percentage of people voting third-party this year might be considerably lower with the lack of any prominent third-party challenger. One reason for the difference is that the underestimation of Trump’s chance in 2016 led a sizable group of people to shift to third-party candidates as a protest vote. The resulted election outcome might have strengthened party unity by encouraging people to stick to a two-party system. This is likely working against Trump – a major beneficiary of protest votes in 2016.

Of course, we are not ruling out the possibility that Trump could win – we still have two rounds of presidential debates ahead and there are extra complexities this year given potential appointment of a new Supreme Court justice right before elections and likely an unprecedented surge in mail-in ballots. However, we do think Biden stands in a more favourable position to beat Trump than Clinton did in 2016. The bigger risk in our view is a potential delay in vote counting due to more mail-in ballots, resulting in no clear-cut winner on the election night, or even in the following week. With several closely contested states allowing mail-in ballots to arrive later than 3 November and an underfunded Postal Service, such an extended period of uncertainty seems a real possibility.

TAO PAN – Head of AI and Big Data

The Algebris Policy & Research Forum (the "Forum") is a not-for-profit advisory forum, independent of the commercial activities of Algebris (UK) Limited. The Forum is designed to contribute to the promotion of a strong and a balanced European economy, underpinned by a sound financial system and supported by a transparent regulatory and fiscal environment, for the benefit of societies as a whole. The primary focus of the Forum is on publishing expert reports, recommendations on European economic policy issues and sharing expert knowledge and research with the public, governmental and non-governmental institutions.
This document is issued by the Forum and the information and opinions contained in this document are for background purposes only, do not purport to be full or complete, do not constitute investment advice and are entirely independent of the commercial activities of Algebris (UK) Limited. Under no circumstances should any part of this document be construed as an offering or solicitation of any offer of any fund managed by Algebris (UK) Limited. This information does not constitute Investment Research, nor a Research Recommendation. No reliance may be placed for any purpose on the information and opinions contained in this document or their accuracy or completeness. No representation, warranty or undertaking, express or implied, is given as to the accuracy or completeness of the information or opinions contained in this document by the Forum or Algebris (UK) Limited, its directors, employees or affiliates and no liability is accepted by such persons for the accuracy or completeness of any such information or opinions. The distribution of this document may be restricted in certain jurisdictions. The above information is for general guidance only, and it is the responsibility of any person or persons in possession of this document to inform themselves of, and to observe, all applicable laws and regulations of any relevant jurisdiction.
© 2020 Algebris Policy & Research Forum. All Rights Reserved.