AMERICA´S ECONOMIC DATA ARE BECOMING MURKIER / THE ECONOMIST

Labouring on

America’s economic data are becoming murkier

Choked for funds, the Bureau of Labour Statistics is cutting corners

A woman and toddler look over dolls on the shelves at a Walmart Supercenter in Burbank, California, USA. /Photograph: Getty Images

EVERY DAY hundreds of Americans take to the streets to collect prices for the Bureau of Labour Statistics (BLS), the government’s economic-data agency. 

Enumerators in 75 urban areas track fluctuations in the costs of a “market basket” of goods and services, from blood tests at doctors’ offices to rental fees for two-bedroom apartments and cinema tickets. 

The numbers they gather feed into the Consumer Price Index (CPI), a statistical tool used to calculate inflation.

In June BLS revealed that two months earlier it had stopped surveying living expenses in Lincoln, Nebraska, and Provo, Utah. 

This month the bureau scrapped Buffalo, New York. 

Producer price data for 34 industries that go into 350 economic indices—manufacturing for lime, ammunition and metal cookware, for example—will also be gone by August, it announced. 

BLS explained its reason for the cuts in a short statement: “current resources can no longer support the collection effort.”

The pruning is the result of two colliding trends that are afflicting not just BLS but also the Census Bureau and the Bureau of Economic Analysis, its sister statistical agencies in Washington. 

The CPI is one of several big economic datasets, many of which rely on ordinary people responding to their knocks and calls. 

Last century Americans did so readily. 

But between 2000 and 2019 response rates for BLS’s flagship employment survey fell from 93% to 83%, forcing statisticians to ask more people questions to get the same quality data. 

Since the pandemic it has plunged to 67%. 

Meanwhile the amount of money that Congress allocates the agencies fell by 18% in real terms over 20 years, even as labour got pricier. 

The agencies were forced to work harder with less. 

To cut costs BLS moved its headquarters from a grandiose building near the Capitol to a cheaper space with a shared cafeteria in the Maryland suburbs last year. 

Since 2019 its bosses have let administrative staff retire without replacing them; today the bureau reportedly has no secretaries.

“It was already barebones, then DOGE came to town,” says Maurine Haver, the head of a firm that sells economic data to big organisations. 

The Trump administration’s hiring freeze coupled with the firing of probationary workers has left the trio of agencies at roughly 60% capacity, insiders say. 

With fewer workers out in the field statisticians have to do more guesswork, which makes the data hazier. 

When BLS cannot find the price of a specific good that goes into the index—say there are no pears in Atlanta’s grocery stores or not enough enumerators on staff to check for them that month—they impute the price of a similar item (apples) or the same item in a different place (pears in Nashville). 

At the start of the pandemic when stores were shut and in-person pricing was difficult to do, the percentage of imputed commodities and services rose from a monthly average of about 8% to 16%, before reverting back. 

In April and May of this year it surged to 30% (see chart).

The consequences of the crunch are obvious to statistically-minded people on both sides of the political aisle. 

William Beach, who was appointed BLS commissioner by Donald Trump in 2017, likens federal statistics to highways and bridges. 

“They are the knowledge infrastructure that the economy needs to function efficiently,” he says. 

Without them bad economic decisions are made. 

For example, if the indices even slightly understate inflation, social security payouts to millions of Americans could be unduly low. 

If the estimate is too high the Federal Reserve might hike interest rates unnecessarily. 

Alberto Cavallo of Harvard Business School reckons that because the CPI is still moving in tandem with online prices, economists don’t need to panic—yet. 

But unlike highways and bridges, which start to collapse when they decay, it is hard to tell how off-target the estimates are becoming.

BLS is required by Congress to produce core economic indicators, regardless of how messy they are. 

But the cuts that make the topline numbers less reliable can have cascading effects of their own. 

When the bureau stopped disaggregating the wages of different types of truck drivers a few years ago, the Teamsters union complained that it could no longer negotiate precise contracts. 

During the pandemic statisticians learned how important it is to publish the price of industrial diamonds, an abrasive used in manufacturing. 

After pausing collection because supply was dwindling they noticed other cost equations getting scrambled. 

“The diamonds affected prices across the whole industrial horizon, I was so surprised,” says Mr Beach, the commissioner at the time. 

More surveys will soon be cut. 

Data series on the teenage unemployment rate, fatal jobsite injuries and union membership are all on the brink of extinction.

Mr Trump’s budget proposes slashing BLS’s funding by a further $56m, or 8%. 

That would make a dire situation worse. 

Erica Groshen, Barack Obama’s BLS commissioner, reckons that there is little incentive to fix the problem in Washington because “nobody gets elected by saying they’re going to make federal statistics as good as they should be.” 

In the meantime economists are keenly watching prices to see how the president’s tariffs play out. 

In April Mr Trump admitted that his trade policies could hurt consumers in the short term. 

“Maybe the children will have two dolls instead of 30 dolls, you know?” he mused. 

“And maybe the two dolls will cost a couple of bucks more than they would normally.” 

But whether American doll-makers will benefit from less competition with China will be difficult to track. 

The price of producing them is set to disappear from the index this summer.