- The main U.S. tally of employment undercounted the number of jobs lost in the second quarter of 2009, a government report suggested on Tuesday.
- The Labor Department's Quarterly Census of Employment and Wages, based on a nearly universal count of 9.1 million employers from unemployment insurance tax records, showed employment fell by 5.1 percent in the year to June 2009.
In contrast, the closely watched monthly payrolls survey, which polls about 160,000 firms, showed a 4.1 percent decline during the period.
The department, scheduled to publish benchmark revisions to the monthly payrolls series when January data is released in early February, has already estimated the March 2009 level of employment will be revised down by about 824,000.
The regular annual benchmark revision is based on the same data set used to compile the more timely quarterly report.
Analysts said the data released on Wednesday suggested the government continued to overcount employment in its monthly payrolls report in the quarter ended in June.
"So that's a little over 500,000 more jobs lost in the second quarter than we knew, even with the benchmark incorporated on the back of the envelope," said Philippa Dunne, editor of the economic newsletter The Liscio Report.
The payrolls revisions next month are most likely to be concentrated in the period from January to March, when the recession was at its worst. The economy lost 741,000 jobs in January alone, according to the preliminary monthly data series.
The expected sharp downward revision to the payrolls series and the latest data suggesting the miscounting has continued point to a more troubling recession than had been thought.
Analysts blame the over counting of employment on the so-called birth-death model that the department uses in its monthly report to try to estimate how many companies opened or closed, a process that has worked well in the past.
"But during this severe recession, it was certainly not able to adjust to those big job losses. If you are looking at the monthly numbers, the birth-death model assumes basically the same numbers for the month in 2008 and 2009 as they assumed in 2006 and 2007," said Harm Bandholz, an economist at Unicredit Markets and Investment Banking in New York.
The department agrees and has said it is working to understand why its model has apparently failed.
One change under consideration is plugging data into the model more often, perhaps quarterly instead of once a year, so that discrepancies would be apparent sooner.