Drew Angerer / Getty

For Donald Trump’s reelection campaign, many numbers matter: the number of Americans who get sick and perish from the coronavirus. The number of months before the economy begins to reopen and rebound. The number of Americans who lose their health insurance and home after losing their job. For political scientists, one number is of particular interest, and it currently stands at –18.3.

That is an estimate of how fast the economy is collapsing in the second quarter, an annualized rate derived by Mark Zandi of Moody’s Analytics. It is a measure of how much less the American economy is producing as businesses close and joblessness swells and the pandemic kills. And it is one of the strongest predictors of whether an incumbent is likely to win reelection: An additional percentage point of GDP growth translates roughly into an additional percentage point of the vote share. Americans are seeing their economy evaporate at the fastest pace in modern history, and that foretells a Democratic landslide in the fall.

Or perhaps not. The coronavirus pandemic is a black-swan crisis happening to a singular president in an extreme political environment. A number that might signal unavoidable doom for another candidate in another election might not for Trump come November.

As a general point, the economy matters for presidential campaigns—not to the extent that it is the only thing that matters, but it matters more than almost anything else. Research shows that voters seem to care about financial conditions more in the year of an election than in the prior years of a president’s term. And they seem to care more about the direction in which the economy’s headed than about its overall health.

The economy helps explain Jimmy Carter’s loss in 1980 and George H. W. Bush’s loss in 1992, as well as Barack Obama’s victory in 2012. The jobless rate was rising sharply during Carter’s and Bush’s reelection campaigns, with the ravages of stagflation and the fallout from an oil-price shock, respectively, souring voters on the incumbent presidents. Although Obama was presiding over a dismal economy when he fought Mitt Romney in his 2012 reelection bid, things were turning around.

This year’s projected headline numbers look dire for Trump, far worse than the numbers ever were for Carter, Bush, or Obama—far worse than they have looked for any postwar president. The jobless rate is anticipated to jump fivefold or tenfold in a matter of weeks, from its current rate of 3.5 percent. Growth is anticipated to tip down from a 2 or 3 percent annual rate to –18.3 percent, as Zandi estimates, if not lower. Household income is anticipated to collapse in unprecedented terms too. Taken by themselves, these statistics point to midnight in America, and a Reaganesque victory for the Democratic nominee.

“The historical parallels are obviously far from perfect, but voters have seemed pretty willing to punish incumbents for the economic impact of global depressions, droughts and floods, and all manner of conditions beyond any president’s control,” Larry Bartels, a political scientist at Vanderbilt University, wrote in an email. “The economic numbers in the next six months will probably look unlike any we’ve seen in the period covered by most statistical analyses of economic voting, so it would be rash to extrapolate the magnitudes of effects implied by those analyses, but even a muted effect will be far from moot.”

Trump, however, may not fit the usual pattern. Whatever the country has pitched into is not a normal crisis or recession. And the American electorate is behaving differently from how it has behaved in the past.

For one, voters have become far less likely to switch parties or abandon their party’s candidate: Trump has fewer on-the-fence voters to compete for, but also fewer to lose. A majority of states were swing states in the 1960s; just a handful are today. Counties have become locked-in too. In the 1992 election, 1,096 were decided by a single-digit margin and just 93 were decided by a 50-point-or-higher margin. In 2016, those figures were 303 and 1,196, respectively. There are no more swing voters. There is no more purple America.

Political identity, partisan affiliation, and polarization have become stronger forces, and  have started to shape the way Americans understand the country’s financial position. Presidents’ approval ratings have become more static over time: Obama did not get as big of a favorability bump from improving economic conditions as previous presidents did, and Trump polled far below where the pre-crisis jobless rate and GDP numbers would suggest. Recent polls show that if a Republican is in charge, Republicans are more likely to see a good economy and Democrats a bad economy when looking at the same economy; the partisan economic-expectations gap has soared. Polarization seems to have fueled a “dissociation” between favorability ratings and underlying economic conditions, explains the political scientist John Sides of Vanderbilt. That opens up the possibility that Republican voters will stick with Trump despite the deep, deep recession ahead.

Another complicating factor: What the country faces is not just a recession, but a natural disaster and a public-health cataclysm and a government-mobilizing battle, together in extremis. Voters tend to punish presidents for recessions. But they tend to reward them for their leadership during wars and other dangerous calamities: Horrors such as Pearl Harbor and 9/11 caused spikes in presidents’ approval ratings, and in some cases smoothed their way to electoral victory. This is known as the “rally ’round the flag” effect, and Trump seems to be benefiting from it: His approval rating has tracked up to 49 percent, the highest of his presidency.

The $2.2 trillion coronavirus-relief package, soon to be followed by another aid bill, and actions taken by the executive branch to distribute medical equipment and other forms of assistance to states might help Trump too, by allowing him to demonstrate his power to voters as they make up their mind. “Relief spending after disasters helps politicians get reelected,” explains Neil Malhotra, a political-economy expert at Stanford University. Data from 1988 to 2004 show that a doubling of federal relief payments to a given county boosts the incumbent’s share of the vote there by roughly 0.5 percentage points. (Malhotra notes that disaster-prevention spending does not seem to have the same effect.)

Moreover, politicians tend to direct federal dollars toward politically sensitive places during a disaster, to reward loyal constituencies and help turn out new voters. The chance of being covered by a presidential disaster declaration doubles for counties in swing states during election years, work by the political scientist Andrew Reeves of Washington University in St. Louis has shown. Such politicking may be happening currently: “The president knows Florida is so important for his reelection,” one unidentified official told The Washington Post this week. “He pays close attention to what Florida wants.”

The sheer complexity and size of this crisis, not to mention the lack of data, also makes forecasting difficult. The United States has held only 18 presidential elections in the postwar period—too small a sample size to draw iron-clad statistical conclusions. None has involved a cataclysm like this, or a governmental mobilization like this. “It is so much of an outlier,” Sides says. “I’m not sure there’s anything we can say with any confidence.”

Still, in lay terms, bad things are bad. Reelection campaigns become more challenging during downturns. Negative headlines, about deaths and ventilators and ruined businesses and doctors wearing homemade protective gear and historical spikes in jobless claims and ineffectual federal aid, dim voter perceptions of the country’s leadership. The rally-’round-the-flag effect tends to be short-lived, often lasting just a few weeks or months. Nothing might persuade Trump’s strongest supporters to abandon him. But the death, destruction, and economic collapse this election year—millions of Americans will inevitably judge Trump for that.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.