It seems to be true, even if it also seems impossible. Certainly the candidates aren’t buying it. In the past six months Trump has posted 93 tweets lauding his and his party’s management of the economy (last week: “Great Jobs Numbers & Economy , Plus!”). Biden is running on a platform that devotes nearly 7,000 words to how he and his party will rescue the economy from Trump (“take immediate, decisive action to pull the economy out of President Trump’s recession.”) The candidates say they can control the economy’s direction, and millions of voters agree, even as they disagree over who would do it better.

But what if Presidents can’t direct the economy—or, if they can, there’s no telling whether they’ll do so for better or worse?

Princeton economists Alan S. Blinder and Mark W. Watson conducted the definitive study on that question, published in 2016. Its findings are powerful and highly relevant to this year’s contest, and they are surprising in two particular ways.

Surprise No. 1: Researching the period from Truman through Obama’s first term, the researchers found that the President’s party made a huge difference: The economy grew faster under Democrat Presidents than under Republican Presidents by a “startlingly large” margin, they report, “so large, in fact, that it strains credulity.” Annual growth in real GDP was 1.8 percentage points greater under Democrat Presidents on average. (Because it would be almost impossible for Presidents to affect real GDP growth in their first days, the researchers assign the first quarter of each new presidency to that President’s predecessor.) Given that the U.S. hasn’t notched a calendar year of 3% growth since 2005, 1.8 points is a gigantic gap. The researchers also showed that judging Presidents on some measure other than real GDP growth wouldn’t change their big-picture result. Their basic finding, they said, “holds almost regardless of how you define success.”

At this point you may suspect partisan bias, especially when you find that the lead author, Blinder, is a Democrat. He served on President Bill Clinton’s Council of Economic Advisers, was appointed Fed Vice Chairman by Clinton, and advised the presidential campaigns of Al Gore and John Kerry. But before you convict Blinder of putting his thumb on the scale, consider…

Surprise No. 2: Blinder and Watson weren’t entirely sure which factors might account for the wide difference between Democrat and Republican Presidents, but they were quite sure what does not account for it: “Our empirical analysis does not attribute any of the partisan growth gap to fiscal or monetary policy.”

So what’s the explanation, if not the President? Could it be the party that controls Congress? No, say the researchers; they checked. After testing many plausible factors and scores of combinations, the best explanation they could propose is a combination of oil shocks, changes in European growth, and changes in overall productivity often associated with long-term technology advances. Those factors share almost nothing in common except that Presidents exert little if any control over them. The most ambitious claims some Presidents could make, the researchers conclude, is that they benefited from “blends of good policy and good luck.”

While the study ended in 2012, extending it to the present wouldn’t likely change the main results. Average annual GDP growth in Obama’s second term and Trump’s tenure through 2019 was the same: 2.5%. If we include this year’s disastrous first two pandemic-influenced quarters, Trump’s average plunges to -0.6% for his presidency through June, widening the party-based growth gap.

None of this means that passionate partisans of Biden and Trump should stop debating economic policy. Maybe arguing for two centuries has helped fuel America’ extraordinary record of long-term economic growth. After all, even Blinder advised a President and two presidential candidates on economic policy.

And for the majority of voters who deeply believe that economic performance reflects the incumbent President’s influence: Third quarter GDP growth will be announced at 8:30 am on October 29, just five days before the election. Mark your calendar.