How Covid Has Changed the Way Governments Collect Data

Business, government and the social sciences are dependent on good data to provide the basis for analyzing the economy.

Quarantine-driven office closures and working from home imperatives rendered many of the traditional methods of gathering such information inoperable. Creative approaches to data gathering based on technological advances have kept that knowledge flowing. An important question going forward is whether these may optimize output and supplant the old methods. JL


Jo McGinty reports in the Wall Street Journal:

Data sets provide critical information about the economy. Before Covid-19, 65% of the Commodities and Services Survey and 50% of the Housing Survey, both components of the Consumer Price Index, the country’s most important measure of inflation, were done in person. To keep the data current while coping with business closures and stay-at-home orders, agencies adjusted the way they collect information, shifting in-person visits to telephone calls or for the first time culling material from alternative sources such as credit cards and company websites. Creative minds found new ways of keeping their series going or coming up with new products identifying what’s going on in the economy

Numerous sets of government data used by private businesses, public agencies and policy makers are collected in person.

So what happens when a pandemic hits, and government workers can no longer visit homes and businesses to retrieve the information?

Before Covid-19 shut down the economy in March, 65% of the Commodities and Services Survey and 50% of the Housing Survey were done in person, according to Jay Mousa, associate commissioner for field operations at the Bureau of Labor Statistics. Both are components of the Consumer Price Index, the country’s most important measure of inflation.

About 90% of the Producer Price Index, 69% of the initial enrollment of the International Price Program, 51% of the Occupational Requirements Survey and 35% of the initial enrollment of the National Compensation Survey are also typically conducted in person.

The BLS data sets—which are just a sample of those affected by the new coronavirus—provide critical information about the economy, and, in some cases, inform other government programs.

The CPI, for example, helps establish tax brackets, Social Security payments, poverty thresholds and eligibility requirements for food stamps and school lunch programs.

“It’s fair to say the CPI touches the fabric of nearly every social program that has been instituted by the federal government,” said Michael Horrigan, president of the W.E. Upjohn Institute for Employment Research and a 30-year veteran of the BLS who previously ran the agency’s price and unemployment programs.

To keep the data current while coping with temporary business closures and stay-at-home orders, the BLS and other agencies adjusted the way they collect information, in some cases shifting in-person visits to telephone calls or for the first time culling material from alternative sources such as credit cards and company websites.

“This has in a way liberated some of the creative minds in the Census Bureau, the Bureau of Labor Statistics and the Bureau of Economic Analysis to find new ways of either keeping their customary series going despite new challenges or coming up with new products that are directed at identifying what’s going on in the economy in these extraordinary times,” said Ken Simonson, chief economist at Associated General Contractors of America and previously a member of the BLS Data Users and Census Bureau Scientific advisory committees.

This isn’t the first time government agencies have been faced with roadblocks to data collection, but the difference during the pandemic has been the length and scope of the interruption.

“Hurricanes come up the East Coast and knock out a bunch of establishments for a week or two,” Mr. Horrigan said. “The Covid-19 pandemic is obviously more widespread. If a store owner lets you in for two years and then says you can’t come in, what do you do?”

In some instances, a program’s method of data collection didn’t change, but its base of operations did.

The Office of Employment and Unemployment Statistics collects about a quarter of the monthly jobs-report data by phone, with 350 employees placing calls from four regional data-collection centers.

“We had to shift everybody to remote work,” said Nicholas Johnson, a supervisory economist in the office. “We had never done that before.”

Workers were sent home and issued laptops and cellphones to use while working remotely.

“The plan is to return to normal operation at the data-collection centers, but we’ve made a permanent decision to keep all staff telework-ready moving forward,” Mr. Johnson said.

But perhaps the most significant disruption has been to the decennial census, which aims to count every person living in the U.S.

Among other things, the numbers are used to divvy up seats in the U.S. House of Representatives, redraw voting districts and allocate $1.5 trillion of federal money annually.

After an initial period of self-reporting, the Census Bureau expected to complete the head count with enumerators going door to door from May through July.

But the pandemic delayed the operation for three months.

The majority of census offices across the country began follow-up work on Aug. 11, and to make up for lost time, planned to stay in the field through October while also asking for a four-month extension to provide apportionment and redistricting data.

Instead, the agency was told to finish data collection by Sept. 30 and meet its normal deadlines, submitting apportionment numbers by the end of December and redistricting figures by the end of March 2021.

At this point, the self-response rate is at around 64%, although residents can still respond by phone, mail, or online.

“They’re already up against it,” said John Thompson, a former director of the Census Bureau. “If we stop there, the undercount will be enormous.”

Without a full and accurate count, the precision of government surveys whose sample frameworks, population controls and statistical weighting are based on the census, will be diminished in the future.

And that would be a blow to many of the data-collection programs that have otherwise rallied during the pandemic.