What’s behind the revisions to Colorado’s job growth numbers

Like clockwork, midway through the state’s monthly jobs report for July, there is the usual note: “estimates were revised” for the prior month.

The explanation was, as always, also included. In June, there were fewer jobs than originally estimated due to “additional responses from businesses and government agencies since the last published estimates,” according to the Colorado Department of Labor and Employment’s July employment report.

Sometimes the revision adds jobs. But lately, it’s been a reduction. June’s numbers were lower than first reported. The state had lost 2,700 jobs, not 1,500.

That’s not great news for the state, of course, but at least in July, employers added 3,700 nonfarm payroll jobs — a number that will likely get revised in the future as more employers share their July job counts.

The seemingly mundane topic of revisions took a turn for the political earlier this month when President Donald Trump fired Erika McEntarfer, the head of the Bureau of Labor Statistics. McEntarfer was nominated by President Joe Biden and received bipartisan approval in an 86-8 vote last year.

The agency released July’s employment numbers, but also revised preliminary numbers from May and June as more employers shared their updates a bit late. U.S. employers had actually added 33,000 jobs those two months, not 291,000 as first reported. Trump called the numbers “a scam.” He’s already picked someone new to lead the BLS, who wants to change the whole system.

The BLS upheaval left many Colorado economists shaking their heads in dismay.

“The data is very important and I have absolutely no question about the integrity of the data or the people putting it together,” said Bill Craighead, an economist and program director for the UCCS Economic Forum in Colorado Springs. “It’s important to recognize that most of the data we look at is based on surveys. But these are high quality, consistent surveys done by the federal government (and) more than just somebody doing a web poll.”

Just don’t read too carefully into any single month, he added. “These are preliminary and subject to revision.”

“Well This Sucks,” was the title of Denver labor economist Ryan Gedney latest newsletter a few days after McEntarfer’s firing. He went on to explain the process of collecting data, why the big revisions weren’t a surprise (at least for those who realize policy changes on tariffs, immigration and federal layoffs will impact jobs) and “certainly don’t reflect any efforts to manipulate or distort.”

Yes, the job is a political appointment but the role of the BLS is apolitical, Gedney said. It measures “labor market activity, working conditions, price changes, and productivity in the U.S. economy to support public and private decision making,” he wrote. “Put another way, if you were to ask the BLS if they see a glass as half full or half empty, they would reply with, ‘we see an 8-ounce glass containing four ounces.’”

Economists know and expect monthly reports to be preliminary, and will be updated not just once, but possibly multiple times and several months or even years in the future.

But the initial high job-growth numbers in May and June didn’t seem right, recalls Phyllis Resnick, executive director and lead economist at the Colorado Futures Center, which is part of Colorado State University.

“I was thinking, really? That’s an awful lot of resilience in the economy in face of very erratic policy and a lot of policy uncertainty with the tariffs going on and off, the impact of the federal labor force and the effects of deportation,” she said. “There were other indications that things were slowing so it was a little bit surprising that the numbers were still so robust.”

Getting fresh data each month about jobs, unemployment and consumer prices provides a glimpse into the current economy and where it might be headed and helps her do her job helping businesses better understand what’s ahead, she said. More comprehensive data is available, but often months delayed and less helpful about what’s happening today or might happen tomorrow.

“I’ve worked in the developing world for years and we’re the envy of the world because we have such timely information,” Resnick said. “But the trade-off is that you don’t get all the information every month on a timely enough basis to have a hard and fast estimate. (The data) lets us keep our finger on the pulse of what’s happening. But we also know that for two months subsequent to the data release, they are subject to revisions as more of those surveys trickle in.”

Where the job data comes from

The BLS relies on surveys of consumers and employers to calculate how many people are working or not, and how many jobs are lost or created.

On who’s working, the Current Population Survey, commonly called the household survey, asks questions to about 60,000 U.S. households each month. Are working-age adults aged 16 and older employed, unemployed but looking for a job, or have they stopped working (like retirees)? This will include counting contractors, gig workers and those still looking for work but no longer getting unemployment benefits. This is what’s used to size up the area’s labor force and the unemployment rate.

To figure out how many jobs were added, the Current Employment Statistics program, commonly called the business establishment survey, asks employers about their nonfarm jobs, worker hours and their earnings. Each month, CES surveys approximately 121,000 businesses and government agencies, which represents about 631,000 worksites.

But the problem with any survey, especially a monthly one, isn’t just getting responses quickly, but getting folks to respond, period. Response rates have been in decline. In 2015, CPS had a nearly 90% response rate. As of last month, it was down to 67%, according to this BLS chart.

Meanwhile, the business survey, or CES, had a response rate of 61% in 2015 and had dropped to 42.6% in March, according to another BLS chart.

The BLS would prefer a higher response rate, but monthly surveys cost money and there was already limited staff before this year’s rounds of federal layoffs.

That’s the similar challenge for the monthly inflation report. Colorado doesn’t have its own. The BLS just focuses on Denver’s rate, which is shared (below), every other month.

When we asked the BLS in 2022 why it doesn’t provide a statewide number, a BLS economist responded, “We don’t have the budget for it.”

Colorado survey sizes and why revisions are important

In Colorado, survey sizes are much smaller. About 2,000 businesses covering 9,700 establishments are surveyed each month, while about 800 Colorado households are asked about their job status, said Tim Wonhof, director of the state’s Labor Market Information at the labor department.

The household survey response rate in Colorado has been around 69% this year, he said. A decade ago, it was closer to 90%.

Timothy Wonhof, program manager and economist at the Colorado Department of Labor and Employment building on Oct. 29, 2024. (Tamara Chuang, The Colorado Sun)

He doesn’t know what the business response rates are for the state but, he said, “Overall response rates for both surveys have been declining.”

When looking at small cities and rural towns especially, proceed with caution, Resnick said. Fewer households are surveyed, if any at all. Every county, though, still gets monthly job numbers, like an unemployment rate.

“In some of those counties, they’re going off just two or three calls. There’s a lot of volatility in the numbers,” Resnick said. “But this is the importance of using the same process month after month. At least you’re not introducing more volatility by changing what you’re doing.”

Late responses always happen. And the BLS does try to impute what the response could be based on the pattern of the employer’s past. And when companies respond late, there could be a change in their hiring, so late responses are valuable, too, even if it causes revisions.

“As Erica Groshen, former BLS commissioner, would say ‘revisions are a feature, not a bug.’ In the first month of collection, about 60-70% of firms have reported their data and that data, plus some imputation of non-responding firms, supports ‘preliminary estimates.’ By the third month, the collection rate is typically over 90%; those additional responses inform revisions (i.e. revisions require less imputation),” Wonhoff said in an email.

The BLS makes clear, he said, the monthly report is preliminary. Two revisions come within three months. Another big revision is once a year after more comprehensive data from the Quarterly Census of Employment & Wages is released.

Employers have to respond to QCEW because that’s what guides how much they pay into unemployment. But the data can be six months delayed, which really hurts forecasting especially if someone’s trying to figure out whether new trade, immigration or other economic policies are hurting or helping before it’s too late.

Maybe the political attention will help Americans and politicians better understand monthly job data. The first release is always preliminary.

“In economic data, there is always a trade-off between accuracy and timeliness. For those who place more value on the accuracy of the data, they are free to ignore the preliminary estimates and wait three months for the final values to get a more accurate picture,” Wonhof said. “Those running the CES program however, based on rigorous research, have decided that there is enough useful information in the preliminary figures to warrant releasing them on a timely basis.”

Take the reader poll

You’ve just read about why revisions are in the news. Does it matter? Take the What’s Working reader poll and help us better report on what Coloradans care about when it comes to the economy. Thanks!

➔ The poll >> bit.ly/WWrevisions

July jobs report: 4.5% unemployment rate

The state’s job situation is looking up, at least based on the preliminary report out Friday.

Colorado added 3,700 nonfarm payroll jobs, while the number of the unemployed workers dropped 4.9% to 148,100. The size of the labor force shrunk a bit, down to 3,280,300 folks who are employed or looking for work.

That translates to the state’s unemployment rate falling two-tenths of a percentage point from June to 4.5% in July, which leaves it still higher than the nation’s rate of 4.2%.

Bill Craighead, program director at the University of Colorado, Colorado Springs Economic Forum, gives an economic update of the region during the UCCS Economic Forum on Sept. 26, 2024. (Chuck Bigger, for UCCS Economic Forum)

Craighead, with the UCCS Economic Forum, called the data “encouraging — keeping in mind that it’s preliminary and subject to revision.”

But, he added, “It’s been unusual for Colorado’s unemployment rate to be above the national rate, so it was nice to see it come down, though it was for mixed reasons — some employment gains but also a small decrease in labor force participation.”

Since Craighead works in Colorado Springs, he also shared what happened locally. The metro area added 900 jobs in July. But June numbers were revised down to 500, from the preliminary 1,600.

“One way I deal with the statistical ‘noise’ is to look at the 3-month average increase, which is now 767 — that’s pretty good,” he said.

More:

• In July, the leisure and hospitality industry added the most jobs, at 6,600, while the trade-transportation and utilities sector lost the most, down 4,000.
• All of Colorado’s metro areas saw unemployment rates fall. Fort Collins had the lowest, at 3.8% in July. The nonadjusted rate compares with the state’s nonadjusted rate at 3.9%. Here are the other metro areas in July, compared to June:
• Boulder, to 4%, from June’s 4.5%
• Colorado Springs, to 4%, from 4.5%
• Denver, 3.9%, from 4.4%
• Fort Collins, 3.8%, from 4.3%
• Grand Junction, 4.1%, from 4.7%
• Greeley, 4.3%, from 4.9%
• Pueblo, 5.6%, from 6.1%

>> View Colorado’s July employment report

Sun economy stories you may have missed

➔ Frequently asked questions — and misunderstandings — about Colorado’s special session to close a nearly $1B budget hole. There’s a lot of confusion out there about why the hole exists, how the state budget works and what’s going to be done to address the gap >> Read story

➔ Colorado approves up to $14 million to plug oil wells before they end up on state’s “orphan list.” The 142 low-producing wells emit “disproportionately high levels of methane emissions,” the state Energy and Carbon Management Commission says >> Read story

➔ $280 million at stake if Colorado cities don’t start complying with housing policies. In a new executive order, local governments will not be eligible for $280 million in grants unless they comply with a range of new housing laws >> Read story

The signatures on works shown in the Federal Prisoner Art Exhibit are obscured, but patrons to the show identified six works created by domestic terrorist Terry Nichols are exhibited at the Federal Prisoner Art Exhibit “Art In Isolation; Creating Space.” (Mike Sweeney, Special to The Colorado Sun)

➔ Terry Nichols and other inmates at Colorado’s notorious Supermax prison are making — and selling — art. Selling the work helps fund restitution. >> Read story

➔ Many Colorado teachers must spend more than 40% of their income on rent. A new report takes a closer look at how some districts are stepping into the role of landlord as scarce housing deters many teachers >> Read story

➔ Colorado tax revenue was especially susceptible to changes in the “big, beautiful bill.” Here’s why. >> Read story

Other working bits

➔ Denver inflation up a bit to 2.1% in July. After an uptick in May, consumer prices for the Denver area fell one-tenth of a percentage point in two months to 2.1% in July. That’s higher than 1.9% a year ago, but lower than July 2023, when the annual inflation rate was 4.5%. Food and energy costs fell in Denver, according to the Bureau of Labor Statistics. >> See Denver CPI

➔ Denver tops city list declining restaurant transactions. The summer restaurant report from Toast, which provides digital payment systems, found that 24 states saw transactions this summer grow. But 13 saw declines, including Colorado, which saw same-store summer transactions fall 2%. But Denver, the report points out, experienced the largest decline among cities, down 5%. >> Toast report

➔ Job fair for Denver-area veterans Aug. 21. DAV and RecruitMilitary are hosting a job fair Aug. 21 at the Gaylord Rockies Resort and Convention Center in Aurora. About 50 employers are participating. The four-hour event starts at 11 a.m. >> Details

Got some economic news or business bits Coloradans should know? Tell us: cosun.co/heyww

Thanks for sticking with me for this week’s report. Share your 2 cents on how the economy is keeping you down or helping you up at cosun.co/heyww. ~ tamara

Miss a column? Catch up:

What’s Working is a Colorado Sun column about surviving in today’s economy. Email tamara@coloradosun.com with stories, tips or questions. Read the archive, ask a question at cosun.co/heyww and don’t miss the next one by signing up at coloradosun.com/getww.

Support this free newsletter and become a Colorado Sun member: coloradosun.com/join

Corrections & Clarifications

Notice something wrong? The Colorado Sun has an ethical responsibility to fix all factual errors. Request a correction by emailing corrections@coloradosun.com.

Type of Story: Analysis

Based on factual reporting, although it incorporates the expertise of the journalist and may offer interpretations and conclusions.

Type of Story: News

Based on facts, either observed and verified directly by the reporter, or reported and verified from knowledgeable sources.