The Bureau of Labor Statistics released its employment report for December. America's seasonally adjusted unemployment rate dropped from 9.8% to 9.4% in December. Is this proof that the household survey is inherently unreliable or that this is a reversal of trend?
Neither!
The unemployment rate has been following a slow downward trend since the fall of 2009. Now looking at the seasonally unadjusted data, we also see it has been falling for five straight months relative to the same month last year (SPLY in postalese.) This trend has been masked by the seasonal adjustment process (see the previous post.)
No doubt analysts will be disappointed by the small increase in jobs (103,000) reported by the establishment survey. Do not give it too much weight. These data will be revised, perhaps drastically in February, leading also to substantial revisions to the GDP series. This will cause the whole history of the recovery to be rewritten.
The good news is that the establishment data may become a more useful real time cyclical indicator. The BLS announced, "Effective with the release of January 2011 data on February 4, 2011, the establishment survey will begin estimating net business birth/death adjustment factors on a quarterly basis, replacing the current practice of estimating the factors annually. This will allow the establishment survey to incorporate information from the Quarterly Census of Employment and Wages into the birth/death adjustment factors as soon as it becomes available and thereby improve the factors. Additional information on this change is available at www.bls.gov/ces/ces_quarterly_birthdeath.pdf."
And that is very good news. Tracking the next cycle with the jobs data will be more accurate.
On a sour note, the BLS will update the household estimates with new population data in February. Historically the gnomes resident in the Postal Square Building have not been time series friendly in their population updates.
