The U.S. unemployment picture looks to be improving but it's increasingly being clouded by shoddy data, a problem that seems to be getting worse as the pandemic progresses.
What's happening: The number of Americans receiving unemployment benefits rose to 29.2 million for the latest week of data, the Department of Labor announced Thursday.
- But the increase of 2.2 million people to the rolls was largely the result of 2.3 million people being added to the Pandemic Unemployment Assistance program in just the state of California.
- California's total PUA additions for one week nearly doubled its total number of recipients and accounted for one out of six people receiving PUA benefits nationwide.
Reality check: "The unemployment data are just incredibly problematic," Heidi Shierholz, a former chief economist to the U.S. Secretary of Labor who now serves as senior economist and director of policy at EPI, tells Axios.
- "When I talk about how bad these data are — and they are very bad — it’s all symptomatic of ... the fact that we have disinvested in our unemployment agencies for four decades and now we are expecting them to deal with thousandfold increases in demand for their services."
- "We set them up to fail by not investing in them in a way that means we don’t have an agile system that can quickly absorb and be flexible to deal with something like this."
Between the lines: Shierholz adds that many of the nation's unemployment systems are running on a 60-year-old computer program known as COBOL that in addition to crashing, often over- and undercounts recipients.
The big picture: While the technology is a problem and the labor market is slowly improving, "we are still in a historically bad situation," Shierholz says.
- "It’s just at absolute crisis levels. We are slowly crawling out of a hole, but it’s a giant hole."