By G C Mays
The Employment Situation Report released by the BLS indicated that non-farm payrolls increased by 120,000 in November. However the household survey indicated that employment increased by 278,000. In the last four months the household survey has shown employment growth of 1.28 million while the payroll survey has been exactly half that at 641,000. Which survey is the more exact measure of employment?
The establishment or payroll survey, as it is commonly referred to, is a sample of about one-third of non-farm payrolls. Using November's numbers this would have represented about 43.9 million people. In contrast, the ADP survey sample size is roughly 23 million people. The payroll survey is straight forward, each payroll a person is on counts as one. This overstates the number of unique people employed because it counts those with multiple jobs multiple times.
The current population or "household" survey is a sample survey of about 60,000 households. It is from this survey that we derive the unemployment rate and more importantly the labor force participation rate and employment to population ratio. This survey also includes W-2 wage earners in its count of the total number employed. However, in addition to W-2 wage earners the household survey also includes the self-employed, which include people who file their earnings using schedule C or freelancers who receive 1099 forms.
Here is where it gets a little tricky, also counted as employed are unpaid family workers, private household workers & people on unpaid leave. People participating in a labor strike are also counted as employed. Unlike the payroll survey, there is no duplication of people. However, including unpaid workers and perhaps even freelancers without a steady income overstates the number of employed.
I decided to take a broader view by comparing annual changes in employment for both surveys and what I found was that since 1975 there has been a gradual trend down in the peak levels of job creation in each successive business cycle.
Although the last four months show a huge discrepancy between the two surveys, year-to-date the difference is not nearly as large. The household survey estimates there are 1.37 million more jobs in 2011 while the payroll survey adds 1.51 million jobs through the first eleven months of the year.
Between 1975 and last month, the U.S. economy created 54.9 million jobs according to the household survey while the payroll survey estimates 52.3 million jobs over the same period. A 2.6 million job discrepancy over an almost 37 year period is not too shabby, considering the differences between who the two surveys count as employed.
What is the impact of the differences?
While the payroll survey overstates the total number of distinct people employed, in my opinion it most accurately estimates the economic impact because it only tracks employment that results in income. The household survey overstates the economic impact of employment because it reports people as employed who may not have earned any income. However, it does accurately capture the number of people engaged in productive activity. If they reported separately the number of people engaged in productive activity but did not earn any income its estimate of the economic impact of employment would be more exact in my view.
Neither survey is perfect but each one fills an important need. Both surveys tend to overstate the number of people employed, albeit in different ways. The household survey allows measurement of the unemployment rate as well as labor force and employment participation. The payroll survey allows us to gauge the economic impact of employment. If the surveys could reverse the long-term downward trend in domestic employment, create jobs and not just measure them we could make some real progress.
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.