Labor Department workforce estimates from the Bureau of Labor Statistics have gone through two major revisions since the early days of the Clinton Administration, just over two decades ago. Along with significant changes to the questions asked of survey respondents on the Current Population Survey (CPS), also known as the Household Survey, six categories defined new alternative measures of labor underutilization for the first time as U1 through U6. The U-series range from intentionally exclusive on the U1 to U3 side, to progressively more inclusive U4 through U6 estimates of the nation's unemployed. Traditional counts ahead of 1994 appear comparable to U5 today, according to analysts and Labor Department insiders, but there is at least some thought that earlier figures could fall between U4 and U5 for the most accurate direct comparison to current unemployment numbers.
Since the Employment & Training Administration (ETA) gathers comprehensive counts from employers' payroll records nationwide, which are supposed to serve as benchmark counts on an annual basis for the Household and Establishment Survey figures released by the Bureau of Labor Statistics (BLS), charts and graphs on Rocky Mountain Perspective use the ETA's more accurate counts. In comparison with the Household Survey, for instance, the 52,000 to 55,000 strategically selected household units which respond monthly from about 60,000 occupied households contacted on a four-month on, eight-month off rotation with a second four months included for year-over-year comparison, the survey's reach would be like counting two or three households per zip code across the country. Of course, the Labor Department weighs urban primary samples units (PSUs) more heavily than rural households, so the reality is that individual zip codes in major cities contribute as many as five households, while many rural areas include just one, or none. That's why the 90% confidence level statisticians regard as 90% likely to be accurate -- or fall within the specified range -- is a rather surprising ± 300,000.
The Establishment Survey is considered more accurate, since 145,000 businesses submit employment data monthly from a sample pool of about 160,000, which shrinks the 90% confidence level to ± 115,000 on the CES. Of course, what's not easily grasped from Labor Department releases is the birth and death cycle which gifts as many as 200,000 or more imaginary jobs on a regular basis, because non-reporting businesses continue to be counted using employment numbers submitted when the business was open. Statisticians' argument for the birth and death cycle goes like this; out of thousands of non-reporting businesses that shutter their establishment, in a "dynamic economy" those businesses are always replaced by new businesses sufficient to make up for job losses from non-reporting employers whose staff continue to be counted as employed for two calendar quarters, or longer. Since 2003, the birth and death cycle has contributed mind-boggling, unsupported numbers to the total nonfarm and total private employment estimates on the Establishment Survey's B-Tables for months not formally considered recessionary by the Labor Department. Mary Bowler and Teresa Morisi are economists in the Office of Employment and Unemployment at the BLS whose observations and insight include some of the underpinnings which help explain what Labor Department data can be counted on to convey. In several educational articles for the DOL, the information officers reference adjustments for population controls in January, and to a lesser extent February, each year and the implementation of the birth and death cycle adjustment/gift on CES figures from 2003 onward as the most significant changes to BLS survey calculations in more than a half century.
The following charts and graphs use ETA nationwide counts first and foremost, given their higher accuracy than what the limited, arguably flawed CPS and CES survey estimates can convey. The ETA's UI employers' payroll records and weekly counts of new registrants/claimants for unemployment, along with Quarterly Census of Employment & Wages (QCEW) counts, provide benchmarks for recalibration of the BLS survey figures on an annual basis..
Since the Employment & Training Administration (ETA) gathers comprehensive counts from employers' payroll records nationwide, which are supposed to serve as benchmark counts on an annual basis for the Household and Establishment Survey figures released by the Bureau of Labor Statistics (BLS), charts and graphs on Rocky Mountain Perspective use the ETA's more accurate counts. In comparison with the Household Survey, for instance, the 52,000 to 55,000 strategically selected household units which respond monthly from about 60,000 occupied households contacted on a four-month on, eight-month off rotation with a second four months included for year-over-year comparison, the survey's reach would be like counting two or three households per zip code across the country. Of course, the Labor Department weighs urban primary samples units (PSUs) more heavily than rural households, so the reality is that individual zip codes in major cities contribute as many as five households, while many rural areas include just one, or none. That's why the 90% confidence level statisticians regard as 90% likely to be accurate -- or fall within the specified range -- is a rather surprising ± 300,000.
The Establishment Survey is considered more accurate, since 145,000 businesses submit employment data monthly from a sample pool of about 160,000, which shrinks the 90% confidence level to ± 115,000 on the CES. Of course, what's not easily grasped from Labor Department releases is the birth and death cycle which gifts as many as 200,000 or more imaginary jobs on a regular basis, because non-reporting businesses continue to be counted using employment numbers submitted when the business was open. Statisticians' argument for the birth and death cycle goes like this; out of thousands of non-reporting businesses that shutter their establishment, in a "dynamic economy" those businesses are always replaced by new businesses sufficient to make up for job losses from non-reporting employers whose staff continue to be counted as employed for two calendar quarters, or longer. Since 2003, the birth and death cycle has contributed mind-boggling, unsupported numbers to the total nonfarm and total private employment estimates on the Establishment Survey's B-Tables for months not formally considered recessionary by the Labor Department. Mary Bowler and Teresa Morisi are economists in the Office of Employment and Unemployment at the BLS whose observations and insight include some of the underpinnings which help explain what Labor Department data can be counted on to convey. In several educational articles for the DOL, the information officers reference adjustments for population controls in January, and to a lesser extent February, each year and the implementation of the birth and death cycle adjustment/gift on CES figures from 2003 onward as the most significant changes to BLS survey calculations in more than a half century.
The following charts and graphs use ETA nationwide counts first and foremost, given their higher accuracy than what the limited, arguably flawed CPS and CES survey estimates can convey. The ETA's UI employers' payroll records and weekly counts of new registrants/claimants for unemployment, along with Quarterly Census of Employment & Wages (QCEW) counts, provide benchmarks for recalibration of the BLS survey figures on an annual basis..
| eta_job_growth_graph_bar___line_graphs_300__2017_jan.feb_v.5.255.pdf |
| excel_summary_ui_claimants_bho_trump_eta_2013-20_w166_rmp_4-2-20.pdf |
| eta_r539cy_hlt_1968-2019_4-2-20.pdf |
