The March Jobs Numbers in Context

For the past couple years, I have been frustrated trying to make sense of the media’s reporting, and the Administration’s touting, of the Bureau of Labor Statistics’ jobs numbers. In addition to conflicting evidence, journalistic laziness, and administrative propaganda, there has been the disjointedness of the reporting: It is one or two data points at a time. That does not help make sense of the numbers. Context is needed; we have to unify to understand.

The result, initially for me as an analyst, and now to share with you, is a simple index that helps make sense of the jobs data. It is based on the strong correlation between the payroll data and the working age population, ages 18-64. Here is a chart of the two data series, with the population data overlaid on the payroll data.

It is reasonable to select the age range of 18 through 64-year olds, and that the two data series are related. Empirically it is compelling. There are a couple asides worth mentioning. The data selected is all of the monthly data available from the BLS (yearly payroll data going back to about 1919 is also available). And the other is there is no generally agreed to definition of “working age.” The BLS’ default definition, e.g., is ages 16-plus. I have used 18 through 64 because it intuitively makes sense and gives a better fit to the payroll data.

Once this relationship is observed, then a job creation index follows. I use the net monthly change in jobs divided by the total monthly working-age population in 1,000’s, and call it JPT (jobs per thousand). This index gives us a tool to easily make comparisons and appeal to standards. Here is a chart of the index from January 1997 to the current jobs estimate of March 2007.

An example of the advantage of appealing to a standard – here, the average – is that the expected number of jobs is 186,000. We get to this result by noting from the data that the total population of 18 through 64-year olds is currently 186 million. Since JPT by definition is one in a thousand and its mean or expected value is 1, we get an expected number of 186,000 net jobs for March. So how would you rate 180,000 jobs? A synonym for average is mediocre. Empirically, the initial net payroll number of 180,000 is mediocre.

Not surprisingly, the bigger picture gets us closer to unity and understanding.

My next post will use the population and payroll data to show (initial estimates) there is a shortfall between 5M and 12M jobs with an approximate $30B loss just in tax revenue.