Processing Your Payment

Please do not leave this page until complete. This can take a few moments.

June 5, 2020

Shaky jobs data muddies COVID-19, Great Recession comparisons

Photo | Wikipedia The U.S. Department of Labor is based in Washington, D.C.

Connecticut was the only state in the country whose April unemployment rate didn’t exceed its peak jobless rate from the last recession, but only if you can trust the numbers.

A newly released analysis of jobless rates by ratings agency Kroll found the distinction: Connecticut’s 7.9% unemployment rate in April remained below the state’s 2008 financial crisis peak rate of 9.3%, recorded in the final two months of 2010.

“This is particularly remarkable given that Connecticut’s unemployment rate was closer to the national average during the Great Recession,” the report said.

Whether or not Connecticut is actually remarkable will be borne out as more state-level economic data is released, including May’s unemployment numbers later this month.

Connecticut’s Department of Labor has called the April numbers unreliable because there were errors in the federal government’s household survey that’s used to calculate unemployment rates. For example, temporarily laid off workers were misclassified, leading to a likely understating of joblessness for the month.

The DOL estimates that Connecticut’s actual unemployment rate for April was in the range of 17.5% -- nearly double its prior peak.

The federal DOL has acknowledged the errors, saying the misclassification affected all states to at least some degree, but there was no immediate correction to the state-level jobless rates, which will be revised later based on additional information.

The Kroll report’s author, William Cox, global head of corporate, financial and government ratings, said in an interview Friday that the survey errors presented a challenge. He said the state-by-state comparison is an attempt to use available data to identify patterns that may be emerging during the ongoing crisis.

"I think we've all learned a lot in this process about what sudden shocks to the economy can do to a lot of these processes and systems that create the data,” Cox said.

For now, the official numbers say that every other state’s unemployment rate in April was above their prior Great Recession peaks, some significantly so.

For example, with unemployment rates of 28.2% and 22.7%, respectively, Nevada and Michigan in April were far above their Great Recession highs of 13.7% and 14.6%.

Kroll said the data “reveal wide differences in the timing and magnitude of unemployment levels on the state level” and indicate that Great Recession unemployment trends may not match what happens in the current recession brought on by the COVID-19 coronavirus.

If the household survey errors understated April unemployment in multiple states, it’s not entirely clear why only Connecticut’s rate was low enough to remain beneath its Great Recession peak.

Andy Condon, director of the Office of Research at the Connecticut Department of Labor.

Connecticut DOL research director Andrew Condon said Friday that he believes the survey errors, in addition to challenges attaining an adequate sample size of households due to surveyors not being able knock on doors like they normally would, were more severe in Connecticut than other states.

“It wasn’t equally dispersed,” Condon said. “We got more than our share of misclassifications in Connecticut, and it doesn’t take that many to affect the numbers.”

He doesn’t think comparing the April numbers across states is particularly useful, as a result.

The bad news is that May’s state-level jobs numbers are expected to suffer from similar problems, Condon said, making it difficult to accurately gauge COVID-19’s impacts on the economy. He expects the problems will be fixed in the coming months, allowing for a more accurate picture to emerge out of year-end job data revisions.

[Read more: America's unemployment rate falls to 13.3% as economy posts surprise job gains]

Besides survey problems, Connecticut’s status as an outlier in the April data could be related to delays in processing jobless claims, Kroll said, noting a mismatch between Connecticut’s claims activity and its official unemployment rate. However, some other states also had those mismatches, such as Kentucky and Georgia, and correlating claims and unemployment rates is not an exact science, the report acknowledged.

“It is unknown at this point whether Connecticut’s relatively modest unemployment will stay intact, or if the gap will close as jobless claim numbers catch up with other states,” the report said. “In KBRA’s opinion, claims data indicate the latter could be true, although differences in data timing and sources make a direct correlation of claims rates to unemployment rates difficult to establish.”

Regardless of the picture in April, Connecticut officials have noted a recent slowdown in weekly unemployment claims, though they remain at record highs.

As Condon wrote in his April jobs report, the looming question now is how many of the state’s lost jobs will come back as public safety permits, and how many of them are gone for good?

Sign up for Enews


Order a PDF