A few days ago, the Washington Post published a long, depressing ticktock of the Trump administration’s execrable attempts to control the coronavirus and “reopen” the country for business, beginning with a shocking but not exactly surprising anecdote about how the president’s economic advisers had abetted the president’s most destructive impulses. Around late March, epidemiological models were showing that the U.S. could face 100,000 to 240,000 deaths. But Donald Trump wanted to rev the economy back up for the sake of his reelection and needed data to justify the decision. Enter the number guys:
A small team led by Kevin Hassett—a former chairman of Trump’s Council of Economic Advisers with no background in infectious diseases—quietly built an econometric model to guide response operations.
Many White House aides interpreted the analysis as predicting that the daily death count would peak in mid-April before dropping off substantially, and that there would be far fewer fatalities than initially foreseen, according to six people briefed on it.
Administration members like Jared Kushner reportedly “embraced” Hassett’s predictions. But as the death toll piled up, it became clear that his model was very, very wrong. One former administration official called it “a catastrophic miss.”
For anybody familiar with Hassett, there was something a bit tragicomic about this whole mess. The man is best known for co-writing Dow 36,000, that totemic relic of tech bubble–era stock market boosterism. He is basically a human incarnation of the Stonks meme who otherwise spent much of his professional life churning out studies predicting that cutting corporate tax rates would lead to improbably huge wage gains for workers. The administration had apparently decided to justify its response to a plague based on the work of someone known above all for making hilariously overoptimistic predictions.
But this all left an open question: What the hell was this model that Hassett had cooked up, anyway?
Like most things about the Trump administration, the answer to that question has turned out to be stunningly dumb.
On Monday, the Post filled in a bit more detail about Hassett’s handiwork, reporting that he had produced a “cubic model” that showed deaths from the coronavirus dropping to near zero by May 15. You didn’t have to be a math geek to realize that sounded silly: People obviously aren’t going to stop dying from this disease in a couple weeks. But the Post’s revelation was met with slack-jawed disbelief from economists, policy analysts, and pretty much anybody who spends time with spreadsheets. A cubic model is not a sophisticated prediction tool. It takes data and uses it to draw a curve, based on an equation you probably learned in high school algebra (y=ax3+bx2+cx+d). You can literally do it with a button on Excel. It’s not a good way to make forecasts in general, since you’re just stuffing numbers in a rigged formula that’s prone to extrapolate small recent trends forward into implausibly dramatic predictions about the future. It’s not an epidemiology tool, and there’s no reason to think it would be a good way to predict COVID-19 deaths. It sounded like Hassett had basically gone Dow 36,000 on the coronavirus crisis, but instead of pointing up, he just pointed down.
Anyway, economist John Voorheis probably summed it up best.
Was it possible something had got lost in translation? That maybe the Post had oversimplified things? Nope. On Tuesday, the Council of Economic Advisers tweeted out the graph showing Hassett’s curve falling to zero by mid-May, much sooner than other models.
When the Post asked Hassett about his modeling exercise, he denied that he’d been trying to predict the coronavirus’ death toll. “I have never, ever said that that’s my projection of what the death count was going to be, and no administration policy has been influenced by my projections,” he said. “It’s an utterly false story that I’ve been a rosy-scenario guy inside the White House.” But the paper’s reporting suggests that, in fact, officials did interpret his curve as a projection. And looking at the CEA graph, it’s not hard to see why: It has a trail of pink dots leading to zero.
Again, none of this is really that out of character for Hassett, who has spent his career making Pollyanna-ish claims justifying conservative policy priorities like tax cuts. But the whole incident seems to have turned into a sort of mental breaking point for the economics profession, which has basically spent the past day dragging the current head of the Council of Economic Advisers, a former University of Chicago professor named Tomas Philipson, on Twitter after he tried to suggest that Hassett wasn’t making a forecast but only “smoothing the data”—which is obviously BS, because you can’t “smooth” data into the future—and insulted Jason Furman, the former Council of Economic Advisers chair under President Barack Obama, who is generally well liked in the field.
The whole thing has been fairly dramatic given that econ Twitter typically has a polite faculty lounge vibe, and speaks to a broader sense in the field that the CEA under Trump has resorted to partisan hackery in a way it hadn’t under previous administrations, both Democratic and Republican.
But the more important point is that the Trump administration appears to have found the dumbest possible math it could use to justify letting thousands of people needlessly die.
Update, May 7, 2020: On Wednesday, the New York Times published an interview with Hassett in which he once again denied that he was using his curve to forecast deaths. “He said that he had never shown the curve to the president or any members of the White House coronavirus task force, and that the White House chief of staff, Mark Meadows, has confirmed to him that no one in the administration has ever used the chart as a forecasting tool.” This would seem to be contradicted by the Washington Post’s reporting, though it’s worth noting that Kushner, while influential in the White House, was not a member of the coronavirus task force. Meanwhile, Hassett confirmed that he drew the curve with the “canned function” on Excel.
