In late March, we talked a bit about what an inverted yield curve has presaged for presidential elections throughout history.
Trump campaigned in part on a promise that his policies would usher in a veritable economic miracle and he’s been at pains lately to explain why things aren’t better than they are considering the sheer grandiosity of his campaign rhetoric. For any other president, explanations wouldn’t be in order because, recent stumbles notwithstanding, the economy is doing fine. But for Trump, “doing fine” isn’t good enough and so, he has to find a scapegoat in Jerome Powell and the Fed.
Because inverted curves typically precede recessions with a lag, those who believe “this time is not different” generally argue that a downturn is likely to show up in 2020 – an election year. Of course you needn’t go by the curve to believe that. This expansion is extremely long in the tooth and tentative signs of economic weakness have popped up stateside at various intervals over the past six months amid a sharp deceleration in the global economy and lingering trade tensions.
As it turns out, the data does in fact suggest that a flatter curve 18 months prior to an election “implies some greater propensity for the curve to foretell a change away from the incumbent President’s party”, Deutsche Bank wrote last month. Obviously, there are all manner of caveats that go along with that analysis, and you can read more in the linked post.
Well, with things starting to heat up (political speaking) ahead of 2020, Goldman is out with an early look at what, one way or another, promises to be a truly historic election.
The bank notes that their clients generally believe the strong economy will help Trump secure a second term. To flesh that out, Goldman looks at four conclusions from their own previous work on how the economy affects election outcomes.
Unsurprisingly, the bank finds that “broad variables like income, employment, and consumption matter more than market-based variables like equity or oil prices” and that “the change is generally more important than the level of most variables, including widely reported levels like the unemployment rate.” Here’s a handy chart which shows the statistical significance of a variety of economic variables:
Goldman also notes that “in some cases the originally reported values are more closely related to election outcomes than the revised values, suggesting that voters might respond to news reports of some indicators, like payroll growth, as much or more than they respond to fundamental effects of a stronger labor market.” That would appear to underscore the notion that voters aren’t all that predisposed to digging too far below the surface if they aren’t personally affected.
The potentially worrisome news for Trump in light of the whole “recession 2020” narrative is that, as Goldman goes on to observe, what happens economically during an election year matters most.
“While some variables are most closely related to the presidential popular vote when measured over a longer horizon, such as real consumption or real disposable income, the strongest statistical relationship with the election result is often with variables measured in Q2 of the election year”, the bank writes, referencing the figure below.
Next, Goldman discusses the incumbency advantage, noting that when you model the vote with economic variables and whether the candidate’s party has “already held two or more terms”, you come away with a 5-6pp advantage for first-term incumbents versus candidates belonging to a party that’s held the presidency for two or more terms.
As far as presidential approval is concerned, Goldman looks at divergences in net approval ratings versus what the collection of economic variables listed above implies. Obviously, this wouldn’t have been a good guide in 2016 (we talked about that at length in the linked post above about the curve), but whatever the case, the bank notes that “if President Trump maintains his -9.6pp net approval rating, combining this approach with our economic forecast suggests that he would win the two-party popular vote by a slim margin.”
That said, the bank rather bluntly states that the polarization along party lines makes approval ratings “potentially less useful” than they once were.
“There has been a growing gap between own-party presidential approval and approval by voters affiliated with the rival party”, the bank writes, stating the obvious, before suggesting that going forward, approval ratings will “still [be] a useful measure” but perhaps more for what they say about the preferences of independent-leaning voters than anything else.
When it comes to the eventual Democratic candidate, Goldman calls predicting who that person will be “nearly impossible” given how crowded the field is. That said, the bank does offer one potentially useful observation, which is basically just that although “the eventual nominee often trails in the polls in the year before the election, they typically have at least some base of support and poll in the double-digits.” Currently, only Biden and Sanders are there, although Kamala and Beto are close.
Crucially, turnout will be key in 2020. Here’s the critical passage from Goldman’s analysis:
This appears to be a more important source of uncertainty than usual at the moment, in light of the surge in turnout in the 2018 midterm election to the highest level since 1914. While there is no historical evidence for the notion that the strong Democratic performance in the midterm election should lead to a win in the upcoming presidential election, the surge in turnout does suggest that something unusual has occurred among the American electorate and raises the uncertainty regarding turnout in 2020.
While the bank admits it’s impossible to determine exactly who voted in 2018, a recent Cooperative Congressional Election Study can be used to gauge the behavior of subsets of the electorate in terms of turnout. On this score, the news isn’t good for Trump. To wit, from Goldman:
Groups that most strongly supported President Trump in 2016 made up a smaller share of total voters in 2018 than they did in 2014. For example, white men over age 65 without a college education supported President Trump in 2016 by a 44pp margin. In the 2014 midterm election, we estimate that this group made up 9.7% of all voters but that in 2018 this share declined to 6.9% (-2.8pp). The pattern is consistent across all white voters over age 40 without a college degree – a key component of President Trump’s base. As shown in Exhibit 11, this constituency voted for President Trump by a 30pp margin in 2016, but their share of the overall midterm vote declined very sharply between the 2014 and 2018 midterm elections. On the other hand, turnout increased for groups that voted firmly against President Trump in 2016, especially among young college-educated as well as non-white voters.
Overall, Goldman’s early read on things is that between the incumbency advantage and the still strong economy, Trump is likely to eke out a second term.
That said, the bank reminds you that very much contrary to Trump’s never-ending attempts to suggest that 2016 was some kind of historic landslide, it was actually anything but (the bank doesn’t frame it that way, probably because they don’t have to – all you have to do is look at the real facts).
“The slim margin that President Trump carried in 2016—the president won Wisconsin, which gave him the 270th electoral vote, by just 0.8pp— suggests the result in 2020 might be quite close once again”, Goldman says, wrapping things up, and reiterating that “the surge in turnout in the 2018 midterm election increases uncertainty further.”
The calculus could (and will) change between now and the vote and if you ask us, there are three factors that could swing things decisively against Trump: i) A downturn in the economy deep enough that the president can’t plausibly attribute it to the myriad scapegoats he’s likely to conjure, ii) A stock market crash that can be reasonably attributed to something he’s said or done, iii) A scenario that sees Beto O’Rourke somehow manage to channel the energy surrounding super-star progressives like AOC and Ilhan Omar while simultaneously distancing himself from their more radical policy ideas and thereby retaining some semblance of centrist appeal (that also assumes he doesn’t fumble or otherwise fall victim to some manner of smear campaign).
Or course all of the above is irrelevant if Trump does what we think he’s likely to do: Influence the actual voting process itself with strong-man tactics and, if he still loses, scream “voter fraud” on the way demanding a recount which he’ll then fix. If you want a blueprint for this, see the Istanbul vote in Turkey.
Read more on election 2020