US Presidential Election Maths
Total Page:16
File Type:pdf, Size:1020Kb
INFOCUS MARKET SNAPSHOT JUNE 2020 US Presidential election maths DISCIPLINED BY NATURE. FLEXIBLE BY DESIGN. HIGHLIGHTED IN THIS PUBLICATION: The icons alongside represent our investment process. Through a disciplined provision of investment policy and security selection at GLOBAL STRATEGIC GLOBAL SECURITY ASSET ALLOCATION SELECTION the global level, regional portfolio management teams have the flexiblility to construct portfolios to meet the specific requirements REGIONAL REGIONAL PORTFOLIO of our clients. ASSET ALLOCATION CONSTRUCTION US PRESIDENTIAL ELECTION MATHS Investor attention over the past few months has, with good reason, been focused on all things related to the Covid-19 pandemic. Understanding the current impact and future consequences of the virus has been of paramount importance. Now that we appear to be over the worst of the crisis in most developed economies, investors are starting to think about other issues such as the impending US presidential election. In this Infocus report Daniel Murray examines the prospects for the two main parties. 2016 revisited Democrat voters appear to have been put off more by Hillary 2016 was a year full of surprises: Donald Trump was not Clinton than potential Republican voters were put off by expected to win the Republican Party nomination, nor was Donald Trump. This is supported by the observation that the he expected to win the race for the White House once the Republican share of the popular vote decreased by about 1% nomination was secured. That he did so was even more between 2012 and 2016 whereas the Democrat share of the surprising given that he won only 46% of the popular vote vote fell by about 3%.1 compared to 48% won by Democrat Party candidate Hillary Clinton. The last time a US president was elected having lost There are six states for which polling has traditionally the popular share of the vote by a larger margin was in 1876. been close between the two main parties. These so-called swing states are: Arizona, Florida, Michigan, North Carolina, Much has been written about the 2016 election and its unusual Pennsylvania and Wisconsin. Hillary Clinton only had to win result and no doubt many factors were at play. Two of the most a couple of these whilst retaining the safe Democrat states important seem to be: to win overall in 2016 but, of these six states, she did not win any. The loss must have been even more painful given that ➡ The relatively high proportion of votes cast for the margin of defeat was only 1% of the vote in four of these neither of the two main parties six states. ➡ The success of Donald Trump in the swing states 2020 forecast The peculiarities of the 2016 vote highlight how challenging 1. Independent share of the vote in US presidential elections it can be to forecast election results. This year will be even more uncertain given the uniqueness of the situation 12 associated with the coronavirus. Whilst there has been 10 a sharp contraction in economic activity in the US there has also been a gargantuan amount of monetary and 8 fiscal stimulus applied, dwarfing even that which followed % 6 the Global Financial Crisis of 2008/09. Furthermore, the Trump administration has become more bellicose in its 4 communications about China in a thinly veiled attempt to 2 appeal to anti-China sentiment that has grown significantly in recent years across the US political spectrum.2 0 1996 2000 2004 2008 2012 2016 Independent share of the popular vote Despite the coronavirus, its seismic impact on economic Source: The American Presidency project at UC Santa Barbara, EFG calculations. Data as at 16 June 2020. activity and the perception by some that the US administration could have handled the situation better, Donald Trump’s The share of the vote that went to independent candidates popularity rating has remained remarkably resilient this in 2016 was the highest since 1996. Moreover, potential year although it has declined a little recently (see Figure 2). 1 US presidential elections are unusual to the extent that they are dominated by only two parties. Whereas the electoral systems of many other countries include multiple parties capable of influencing the result, in the US the two main parties have in effect had a duopoly for at least 100 years. This means that independent candidates are often from fringe parties and therefore lack the resources and credibility to attract a meaningful number of votes. The last independent candidate to make a significant mark on the US presidential election was Ross Perot who stood in both 1992 and 1996, attracting 18.9% and 8.4% of the vote respectively. However, he failed to win any electoral college votes. In the 2016 election Libertarian candidate Gary Johnson won 3.3% of the vote and Green party candidate Jill Stein won 1.1%. 2 See recent Macro Flash Note from 3 June 2020, “China-US relations: it’s complicated”. 2 | June 2020 US PRESIDENTIAL ELECTION MATHS 2. Trump’s popularity rating (7-day moving average) The analysis also controls for whether or not the incumbent President is running for a second term, the number of 56 consecutive terms for which the incumbent party has been in 54 power and for war years. 52 50 3. US presidential elections: Democrat share of the vote, % 48 actual and predicted 46 70 44 65 42 60 40 55 Jan Feb Mar AprMay Jun 2020 Approve Disapprove % 50 Source: fivethirtyeight.com, EFG calculations. Data as at 16 June 2020. 45 40 Aggregating polling data taken from fivethirtyeight.com across 35 47 different pollsters shows that Trump’s approval rating 30 stayed in a range of between about 42 and 44 until late March. 1916 1928 1940 1952 1964 1976 1988 2000 2012 2020 It then briefly rose before settling at around 44 where it stayed Democrat share of two-party popular vote Forecast Democrat share of vote until the end of May. It has declined since then to levels similar Source: Ray Fair, The American Presidency project at UC Santa Barbara, EFG calculations. Data as at 16 June 2020. to those seen in the early part of the year. It is possible that the coronavirus and its associated impact It is hard to say for certain what all these cross currents mean will cause this model to break down for the 2020 election. for November’s election but it is possible to use econometric However, as we don’t know how the model will be affected, techniques to help form an objective statistical view of how it seems reasonable to use its output as a starting point. It to gauge each candidate’s chances. A well-known academic is also worth noting that all elections have unique features paper from 19783 tested a variety of different inputs to see associated with them, despite which the statistical model has how well they predicted the Democrat share of the vote in performed well historically. presidential elections. The results show that it is possible to predict the outcome of presidential elections using a small Taking actual Q1 2020 data and using consensus forecasts number of economic variables. While other factors are of for the second and third quarters produces an annualised course important, the findings suggest that economic factors growth rate in real GDP per capita of around -10.5% for the first are meaningful determinants in explaining election results. three quarters of the year. Using actual data where it exists and consensus forecasts for the second and third quarters The same author has updated the analysis for every election allows us to produce input values for the other two variables. since then. The latest analysis – incorporating the 2016 election Plugging these variables into the equation produces a forecast result – suggests that the economic backdrop can be captured for the Democrat share of the vote of 53%. Plugging in numbers by only three variables: as they were at the beginning of the year, including growth and inflation estimates for 2020 at that time, would have 1. The per capita GDP growth rate in the first three quarters produced an estimated Democrat share of the vote of 46%. of the election year (a higher value reduces the predicted The major difference between the beginning of the year and Democrat share of the vote) now is that growth estimates for 2020 have been revised significantly lower. 2. The average inflation rate over the first 15 quarters of an administration (a higher value increases the predicted The model assumes that there are only two parties so that Democrat share of the vote) the forecast Republican share of the vote is simply 1 minus the Democrat share. In this instance that would be 47% for 3. The number of quarters during an administration in 2020, one percentage point more than Trump received in 2016. which the growth rate of real GDP per capita exceeds However, we need to adjust the 2016 result to take into account 3.2% (a higher value reduces the predicted Democrat the high proportion of independent votes: of the votes cast share of the vote) only for Clinton and Trump, Clinton won 51% and Trump 49%. 3 Fair, Ray C., ”The Effect of Economic Events on Votes for President”, Review of Economics and Statistics, 1978, 60, 159-173 June 2020 | 3 US PRESIDENTIAL ELECTION MATHS For reference the model anticipated that the Democrats would Conclusions win 44% of the vote in 2016 and the Republicans 56%. This is Predicting the outcome of this year’s presidential election interesting because (a) it was quite inaccurate in terms of the is even more challenging than normal given uncertainties actual share achieved – which illustrates the importance of associated with the coronavirus pandemic.