(Bloomberg Businessweek) -- Checking the latest Covid-19 numbers has become a daily ritual in many American households. There’s both solace and alarm to be had from the numbers—one person’s comforting plateau in infection rates is another’s alarm at the lack of testing. Still, there’s agreement that no data set matters more right now. It guides life-and-death policy decisions from where to allocate protective equipment and ventilators to when it might be safe to get back to business.
In that context, the traditional indicators by which we gauge the health of our economy seem suddenly inadequate. From the Federal Reserve to Wall Street, there’s a scramble to find a new high-frequency data to map the destruction in real time and help guide the salvage effort. Which means that one lesson of this crisis ought to be that it’s not just pandemics that you need to prepare better for, it’s the rapid recessions that come with them as well.
When the world’s leading central bankers dialed in to a conference call in early March for the regular bimonthly meeting of the Bank for International Settlements, some of them were already complaining that they were flying blind because of the inevitable lag in economic data.
Consider, for instance, the Bureau of Labor Statistics’ monthly employment report—a highly scrutinized release that often moves markets. The March edition pegged job losses at 710,000, because it sampled data that went only through mid-month, before several U.S. cities and states began mandating shutdowns. A different set of numbers—the weekly unemployment insurance claims—showed more than 17 million people filed to collect benefits between the beginning of March and April 4.
Those numbers are also problematic. From around the country, there have been reports of websites crashing and hotlines ringing busy, as state agencies struggled to cope with the historic deluge of people filing for unemployment benefits. Also, in states including Massachusetts and Oregon, whole segments of the labor force—the self-employed and gig workers—have been told to not file their claims for the time being.
Concerns over the quality of the jobs statistics prompted a group of Federal Reserve Board economists to construct their own employment index using data from payroll services provider ADP LLC, which can churn out daily stats covering 20% of U.S. companies. By its calculations the U.S. lost a net 13 million jobs in the final two weeks of March alone. Yet it also admitted that if a different methodology were used, the job losses would have been as high as 23 million.
Number crunchers at the International Monetary Fund are also having to get creative in how they go about forecasting the contraction in economic output that official statistics will eventually reveal. Their latest projections have the global economy shrinking by 3% this year. To arrive at that figure, they assumed that countries suffering severe outbreaks will lose roughly 8% of working days to stay-at-home orders and other efforts to contain the virus.
For the U.S., that means the economy would be operating on the equivalent of 230 working days rather than the standard 250 or so. Those 20 lost days are deliberately fewer than the days in which containment measures are expected to be in place, as some businesses have found ways to keep operating while shutdowns orders are in effect. (According to the BLS, roughly one-third of Americans can work from home). But who’s to say that it couldn’t be 30 days? Or more?
The unprecedented speed and scale of this recession makes it almost impossible for forecasters to tell us how bad it’s going to get. Some have given up. The Philadelphia Federal Reserve Bank recently suspended publication of its state-level leading economic indices indefinitely because the “extreme impact” of the Covid-19-related job losses had made the exercises unreliable for predicting the next six months, which is what they are designed to do.
Other forecasters are relying heavily on documented historical experience to predict the impact of a freak event. No surprise then that the latest projections by analysts at banks and credit-rating companies span an eye-popping range. Some two-dozen estimates for second-quarter annualized gross domestic product growth collected by Bloomberg since April 10 range from -65% to 0.4%.
Peering into our future might be easier if we knew with certainty what happened in our immediate past. Measuring the GDP of the U.S. is a giant forensic exercise, and in a normal year we’d be resigned that the staff at the Bureau of Economic Analysis wasn’t going to rush it. Having to wait till April 29 to get preliminary estimates for the first three months of 2020 feels specially pointless in a year in which time is no longer measured in quarters but as B.C. and A.C. (before corona and after corona). And early figures for the three-month period everyone really cares about—the one we are in now—won’t come until the end of July, perhaps too late to help policymaking.
Our appetite for more timely data is feeding a fascination with high-frequency measures and what are normally considered alternative indicators. Economists and investors are scrutinizing everything from restaurant, hotel, and airline reservations (all down) to metrics of electricity usage (also down) and credit card spending (yup, down).
Or, put another way, it’s why we all have a little Li Keqiang in us these days.
One of the tidbits we learned from the 2010 WikiLeaks dump of U.S. diplomatic cables was that Li, China’s current premier, doesn’t trust the country’s economic data. Or at least didn’t in March 2007 when, as a provincial leader and rising star in the Chinese Communist Party, he told a visiting U.S. ambassador that China’s GDP figures were “man-made,” hinting that career-driven cadres in the provinces were fudging their numbers to impress their bosses in Beijing.
Li confided that he developed his own gauge based on data he regarded as more reliable, including electricity usage, rail cargo volumes, and bank loans. “All other figures, especially GDP statistics, are ‘for reference only,’ he said smiling,” the ambassador, Clark Randt, reported in his cable. The disclosure led to a rush by news organizations and data providers such as Bloomberg and the Economist to create their own versions of the Li Keqiang Index. Now prodded by the pandemic, many organizations—Bloomberg included—are attempting do the same for the U.S.
It merits saying that alternative data—any data—should always be treated carefully. It can guide bad policy. During the Vietnam War the obsession of Defense Secretary Robert McNamara and other senior U.S. officials with keeping a daily “body count” of the number of enemies killed was widely blamed for a conviction that endured for too long in Washington that the U.S. was winning a bloody war it was in fact losing.
One problem then was that, much as in China, promotion-seeking American officers on the ground were inflating the numbers they were reporting up the chain of command. But there was also a blind faith in the data. What’s now known as the McNamara Fallacy is best described as that fugue state that occurs when decision-makers with their eyes trained on quantitative metrics end up missing the bigger picture.
U.S. policymakers are already giving alternative metrics more weight than they used to. As a giant wave of layoffs began sweeping the country in March, one senior White House official revealed privately that the administration’s conception on the extent of the damage changed following a briefing with credit card companies, which have a real-time window into consumer spending.
Since 2014, the Federal Reserve of Atlanta has been publishing a “GDPNow” index that’s updated constantly with new economic data releases from consumer spending to corporate inventories. But it’s described by its creators only as a “running estimate.” It now comes with a user warning that the measure “does not anticipate the impact of Covid-19 on forthcoming economic reports.” It also hasn’t always been used judiciously. White House economic adviser Larry Kudlow was citing its benign readings into early March as a sign of what he was then forecasting would be a limited impact on the U.S. economy.
The Federal Reserve Bank of New York produces its own Weekly Economic Index built on metrics such as retail sales, unemployment claims, a daily consumer sentiment gauge, steel production, and electricity usage. It too, however, comes with a health warning. Alternative data amount only to “an informative signal of the state of the economy,” its makers caution, and the index itself is only “a parsimonious summary of that signal.”
What we are now going through, though, has demonstrated our need for something less parsimonious. A push into live, or near-live, short-term data—that’s able to track everything, from consumption to the realities of global trade and supply chains, and that’s more reliable and useful in a crisis—ought surely to be a priority for governments around the world. The data may not be good enough to help decision-makers quickly bend the curve of our economic suffering in the current crisis. There’s plenty of people with an interest, though, in making sure that’s not the case next time.
Read more: Risk Manager Is Suddenly a Hot Job in the Coronavirus Era
©2020 Bloomberg L.P.