Recent predictions about the economic impact of the COVID-19 crisis are certainly helping the “dismal science” live up to its name. Last Tuesday, an International Monetary Fund report said that the world is hurtling toward the worst economic downturn since the Great Depression. Economists at big investment banks have warned more or less the same: Goldman Sachs recently predicted that U.S. GDP would shrink by 34 percent in the second quarter of 2020, compared to the first quarter of the year, and Morgan Stanley forecasts that this year, we could see the steepest drop in annual GDP growth since 1946.
But here’s the thing: Economists aren’t especially good at predicting recessions. In this instance, with at least 20 million people out of work in the U.S. and pretty much every country suffering from the pandemic, it seems almost certain that we’re in one. But forecasting the path of a recession isn’t an easy task under the best of circumstances. And we are not living through the best of circumstances right now. “This is not a situation where you can push a button on the computer and out comes a number,” said Jonathan Wright, a professor of economics at Johns Hopkins University. “It’s detective work. And it will mostly be wildly wrong.”
And we also don’t know how it will be wrong. The forecasters are pretty much in agreement that the next months are going to be full of economic pain — but there’s a lot less consensus about how quickly the economy will bounce back. Goldman Sachs, for example, is predicting an “unprecedented” recovery in the second half of the year, once businesses start to reopen. The chief of the San Francisco Federal Reserve Bank, meanwhile, says she’s expecting a much more gradual return to positive economic growth. If anything, a recent study suggests that forecasters tend to be too optimistic about when recession recoveries will begin, which means the return to a normal economy could be slower and bumpier than many economists are currently predicting.
Look at recession predictions for the past year, and you’ll start to understand why some economists raise an eyebrow at some of their colleagues’ efforts to play Cassandra. After a flurry of ominous warnings about an impending recession last summer, most forecasters were predicting a strong year for the U.S. economy. Gregory Daco, the chief U.S. economist at Oxford Economics, told me a story about a presentation he gave at a conference on the global economic outlook back in February. At that moment, the novel coronavirus mostly hadn’t reached U.S. shores, but Daco had been keeping an eye on the virus in China and told the assembled crowd of economists that in a worst-case scenario, U.S. GDP growth could fall as low as 0 percent in 2020. Some people in the audience were not pleased. “It got back to me that they thought my estimate was way too pessimistic — there was no way we would be in an environment where the U.S. economy wouldn’t grow because of the virus,” Daco said.
It’s amazing what two months can do. Now, with much of the global economy on lockdown, that prediction feels outlandishly optimistic. Daco’s latest estimate is that the economy will contract by 4 percent over the course of 2020 — assuming a healthy rebound in the last few months of the year.
Largely failing to predict the economic impact of the COVID-19 crisis, even when it was already wreaking havoc on one of the world’s most populous countries, might seem like a huge mistake. But it’s actually very normal for recessions to catch forecasters by surprise. Forecasters didn’t just fail to predict the global financial crisis of 2008 — the recession had been rolling along for almost a year before it officially got the label. One study published in 2018 looked at more than 150 recessions across the globe and found that only a handful were successfully predicted by economists. Apparently, professional economic forecasters are an unusually sunny folk — or at least their predictions are. Research has found that economists consistently overestimate economic growth.
All of these challenges make sense, when you think about it. “It’s not as if recessions are black swan events and we should never be prepared for them, but it’s true that they are unexpected and hard to predict,” said Prakash Loungani, an economist at the IMF who studies recession predictions. Forecasting models also rely heavily on historical data, and recessions are rare. Since 1970, the U.S. has only seen seven. And the very nature of recessions make them hard to predict based on the past. “No model that’s used in normal times will forecast a recession because by definition, it’s a break from normal times,” said Claudia Sahm, the director of macroeconomic policy at the left-leaning Washington Center for Equitable Growth and a former economist with the Federal Reserve. “So you have to use your judgment and look at indicators, like the unemployment rate, for clues about what’s going on.”
To complicate matters further, each recession is unique, starting with the economic conditions that triggered it. That makes relying on history an even trickier business. “The last recession was connected to a financial crisis and this one is a public health crisis,” said Tara Sinclair, a professor of economics and international affairs at George Washington University. “This is one of the problems in forecasting recessions — the people who are experts on the issues underlying the recession this time are not necessarily the same people who were experts last time.”
Once a recession begins, similar dynamics usually start to emerge — consumer confidence falls, the stock market tumbles, the unemployment rate goes up. But assumptions about how recessions will unfold or how long they’ll last may not hold up from one crisis to the next, and it can be dangerous to rely on them. A recent study by Loungani found that in 436 recessions since 1990, IMF forecasters mostly predicted a rapid recovery. (Consensus forecasters had a similar track record, but Loungani had a smaller number of predictions to analyze.) In many of those recessions, to be fair, the bounce back was pretty quick. But overall, Loungani found that forecasters did not do a good job of predicting which recessions would drag on for more than a year — including the global financial crisis of 2008, where economists predicted that a recovery was coming long before indicators like the unemployment rate returned to their pre-recession levels. That misplaced optimism about a quick return to normal may have actually prolonged the recession, according to research by Loungani and others, by spurring politicians to end fiscal stimulus measures before the recovery was fully in swing.
The COVID-19 pandemic is making economists’ jobs even harder by taking some of the usual sources of uncertainty and turning them upside down. Unlike in past economic downturns, there is pretty much no debate about whether we’re in a recession. That’s in part because the recession was triggered by the government as it sought to respond to the pandemic. “We’re aware of the economic problem almost immediately because the public health measures we’re taking created it,” Wright said. That’s helpful in one sense, because it allows policymakers to respond quickly. “But it seems odd to say it’s a good thing,” he added. “The only reason we know we’re in a recession so quickly is because the shock is just enormous.”
Knowing that we’re in a recession also doesn’t make forecasters’ job easier. In fact, the trajectory of this crisis is even harder to predict because it’s so dependent on external factors, like when shelter-in-place orders will be lifted. Wright said it’s possible to get a reasonably accurate estimate of the short-term economic damage from indicators that update quickly, like weekly unemployment claims, electricity consumption or output from other sectors of the economy, such as industrial production or retail sales. But there’s a lot we just don’t know — for instance, how many small businesses will be able to reopen when the lockdowns end. Their ability to start rehiring workers will have a big impact on how quickly the economy can recover. And even rapidly updating statistics, like those unemployment claims, can’t keep up with the breakneck pace of the economic crisis.
Daco, whose company has started issuing biweekly forecasts for its clients, told me that he’s simply expecting to be wrong a lot. “As forecasters, we have to approach this with a heavy dose of humility, because we just don’t know how this will evolve,” he said. “One day we could be assuming a 10-12 week lockdown, but the next day we could be looking at a lockdown that lasts through September and you’ve got to throw yesterday’s assumptions out the window.”
That doesn’t mean the forecasts are useless. But Sinclair told me it’s better to think of them as a range of possible futures, rather than a reliable vision of what will come to be. Right now, she said, there are too many unknowns, starting with the question that’s at the top of everyone’s mind: When will it be safe to leave our homes again?