Last Friday's employment report befuddled everyone. Suddenly, the unemployment rate dropped to 9%, which should be good news. But the household employment survey, which includes self-employment and is the data upon which the unemployment rate calculation is based, underwent an annual revision of prior data that was so large that even the Bureau of Labor Statistics stated that December 2010 and January 2011 were not comparable.

January 2010 and January 2011 are far enough apart that they are comparable. The unemployment rate in January 2010 was 9.7%, and it has now dropped to 9%. The number of employed people has increased by only 0.6%., or 812,000. Since that time, more than 2 million people left or have not entered the labor force. Basically, for every jobholder in the labor force being added, two people are leaving the labor force. One of the key employment measures is the ratio of employment to population. That was 58.5% in January 2011 and is now 58.4%, or essentially no change. The labor participation rate is continuing its decline, and is now 64.2%. Even though the unemployment rate indicates improvement, the underlying data are still showing marked deterioration.

The BLS revision to the payroll employment historical data was also large. It was so large that the revisions seem stratospheric. Through 2010, the BLS had advised users of its data that the revisions were coming, but it's not until you actually see the data that you can understand their nature. The BLS had overestimated payroll employment by more than 500,000. Once that was removed, it meant that employment since April 2009 was essentially unchanged. Since January 2010, there are 984,000 more people on payrolls. Since January 2008, just after the official date of the recession's start, payroll employment is down -7.7 million.