Monday, February 10, 2014

Hussman Weekly Market Comment: Double Trouble

Link to: Double Trouble
Economic Notes 
One of the seemingly confusing aspects of Friday’s employment report was the increase of 638,000 jobs in the household survey, which contrasted with the rather disappointing 113,000 jobs in the more widely followed payroll survey. What many of the talking heads called a troubling difference can be explained by a footnote in the report from the Bureau of Labor Statistics: “household survey data for January 2014 reflect updated population estimates.” The civilian labor force itself was revised higher by 523,000 individuals. 
Here’s how these revisions work (from Business Statistics of the United States, Strawser, 2012): “Official [current population survey] data are characterized by periodic discontinuities, which occur when benchmarks for Census measures of the total population are introduced. These updates take place in a single month – usually January – and the official data for previous months are typically not modified to provide a smooth transition… Such discontinuities occur throughout the history of the series.” 
One of the more notable examples of this is was in January 2000, when the household employment figure jumped by 2,036,000 jobs, while the civilian labor force jumped by 2,090,000. By comparison, the total non-farm payroll figure (establishment survey) increased only by 233,000. The nearly nine-fold difference between the household and establishment figures wasn’t some gross aberration or cause for alarm in the field of economic analysis. It simply reflected standard practice in adjusting the household data for changes in estimated population. 
Similarly, in January 2003, the household employment figure surged by 991,000 jobs, while the civilian labor force figure jumped by 871,000. In contrast, the total non-farm payroll figure increased by only 89,000 jobs. This simply reflected standard practice. There are certainly times when true inconsistencies in payroll figures can be informative, but where January benchmark revisions are concerned, there’s really nothing out of the ordinary here.