This past Friday's unemployment report has caused a great deal of confusion among businesspeople and (always-confused) political classes. Politically-charged pre-election periods make discerning the nature of economic data more difficult. I've dug into the data and these are the inconsistencies in the report that have caused the confusion and offer some perspective. The report is internally inconsistent, and conspiracy theories should be expected but not necessarily believed. There are flaws in the unemployment report, just like all data collection processes. These are my notes that will become a couple of slides in my Monday presentation at Graph Expo for which there is still room (sign up here) and we promise the legendary McCormick Place coffee and bagels and other high-carb essentials :)

The latest unemployment report is always at http://bls.gov/news.release/empsit.nr0.htm and has links to all of their tables at the bottom of the page

•Unemployment rate dropped from 8.1% to 7.8%

•Payroll survey is not used to calculate unemployment rate; added only 114,000 jobs

–Goods-producing industry employment went down

–Transportation and warehousing went up... so there was more employment to store and ship goods that weren't being made

•Household survey used to calculate rate

–Added 873K jobs, or 659K self-employed or 1099 (non-W-2) jobs

–Bulk of job increase was among teenagers, black, and “some high school” classifications; these are typically late cycle growth categories after economic growth has become sustained

–“U-6” broad measure of unemployment stayed exactly the same, 14.7% when it should have reflected the improvement of the overall rate

•Change in the unemployment rate is not historically consistent: In periods of economic growth, the unemployment rate gets worse for about six months as optimistic and motivated workers flood the market faster than jobs are created

•BLS has been constantly changing its seasonal adjustment factors that are not designed for prolonged sluggish economic periods; one month’s factor is not created by the same technique as the prior month’s; BLS is flailing about seeking the right seasonal adjustment methodology

•BLS is also secretive and does not release raw data for independent analysis, which promotes “economic conspiracy” propositions; more transparency would be helpful. While the data are not released, the methodology is well documented.

•Report was not consistent with other independent data and data from other agencies

•Government economic statistics are notoriously bad in the short term, and become better after years of revision; unemployment report is the closest “real-time” report available, but is revised monthly, annually, and then subject to annual three-year revision cycle. This means that accuracy comes at a time appropriate for historians, and not in a time-frame needed for policy decisions.

•It is naïve to expect data based on estimates of estimates and seasonal adjustments in a mostly free and dynamic economy to have Census-style accuracy; not even the Census has the certitude that is attributed to it.

Since the start of the recession in December 2007, and despite the recovery since June 2009, there are 3.2 million fewer employed, and the labor force has grown only by 1 million workers (it should grow by about 3 million per year).  The labor participation rate, that is workers as a percentage of the possible work force, has fallen from 66% to 63.6%, which is at about 30 year lows. The employment as a percentage to total population has fallen from 62.7% to 58.7%. Despite improvements that appear as headlines that cite the unemployment rate only, the deeper looks at employment data still indicate retrograde labor market conditions.

# # #