The January 2011 monthly unemployment figures are out. The official unemployment rate decreased to 9.0% and the total jobs gained were 36,000. Total private jobs came in at +50,000, in stark contrast to the 187,000 ADP reported for January. You might ask how the unemployment rate could drop when there were not enough jobs added to even handle U.S. population growth? This month is funky. The BLS changed population ratios, an annual revision, but as a result makes month to month change comparisons invalid. Why is explained below.
The BLS did two major adjustments, the first being yearly revisions to nonfarm payrolls, or the actual number of jobs and the second being to population. For The entire year of 2010, nonfarm payrolls were adjusted downward by -215,000. Now this is the CES survey, which extrapolates the total number of jobs from direct survey data. Here are the details of the CES survey, from the BLS:
BLS collects these data each month from the payroll records of a sample of nonagricultural business establishments. The sample includes about 140,000 businesses and government agencies representing approximately 410,000 worksites and is drawn from a sampling frame of roughly 8.9 million unemployment insurance tax accounts.
So, in terms of counting the actual number of jobs created in January, we can assume these are reasonably valid, in spite of an annual revision from the previous months of -215,000 in total nonfarm payroll jobs.
The population adjustments are where things get weird. This is the CPS survey, which is a sampling of 60,000 households, then that data is extrapolated to total population. In turn those CPS metrics are ratios, based on assumed population growth. Now, even though the 2010 Census data is out, the BLS is basing population changes from the 2000 Census and, won't be modifying the Household survey data for the unemployment report with 2010 Census data....until 2012.
The actual population totals were updated by the Census in terms of actual growth projections....but....that is running from a Census 2000 baseline, not 2010. To add to that, these adjustments affect all previous months, not just December and January, even though the adjustment occurs between the December to January in the data. Get that? In other words you have a glitch in the data progression timeline, so comparing December to January includes that glitch or adjustment, even though that adjustment represents months of cumulative population data errors going backwards over time. Therefore, one cannot compare the change from December to January in these months for total CPS amount changes and have it be valid. That change from December to January includes the glitch or the adjustment number. The Unemployment Rate is a Ratio and all things are derived from the estimated United States population. Here are the adjustments listed by the Census:
The adjustment decreased the estimated size of the civilian noninstitutional population in December by 347,000, the civilian labor force by 504,000, and employment by 472,000
Below is an example of the glitch from the civilian non-institutional population. The graph represents the percent change, monthly. Notice how the increase is pretty much 200,000, except once yearly, from December to January, where you get those huge changes. That's our yearly population adjustment, or glitch, or data discontinuity.
So, with that, let's look at what we know is valid in trying to compare January 2011 to December 2010.
Below is the nonfarm payroll, the total number of jobs, seasonally adjusted. Since the start of the great recession, declared by the NBER to be December 2007, the United States has officially lost 7.731 million jobs. This is with the new adjustments from past months for nonfarm payrolls.
Below is a running tally of how many official jobs permanently lost since the January 2008, 1 month after the official start of the past recession (recall the private NBER has declared the recession over!). This is a horrific tally and notice this isn't taking into account increased population growth from December 2007.
How can the unemployment rate decrease so dramatically when not enough jobs were added to even cover population growth? Here is where those population adjustments kick in, so we can only give the current ratios for any sort of data validity.
The civilian labor force was 153,186,000 and the new civilian population for January is now 238,704,000. Of those still in the civilian labor force, employed, was 139,000,00, yet those unemployed was 13,863,000. Those not in the labor force was 85,000,000. The employment to population ratio is 58.4%.
The civilian non-institutional population are those 16 years or older not locked up somewhere or not in the military or so sick and disabled they are in a nursing home and so on. Not in the labor force are those of the civilian non-institutional population who are not counted as looking for work or having a job. The BLS gave us a handy chart to show the change as reported and what the change would have been in the third column, if the population adjustments were removed from the January reported CPS numbers, so one could compare Apples to Apples, the fruit being December 2010 to January 2011. Still, although the fruit is now the same in the below chart, it's still poison.
Now you say, ok, where did the the 589,000 employed people come from if you take away the population adjustments? Well, frankly they didn't come from anywhere, remember the CPS survey is extrapolated, so because the overall population is lower and now adjusted, those 589,000 are imaginary people now, based on faulty overall population numbers. Weird huh?
What this tells us is to focus on the CES survey this month, which also was adjusted and maybe write your Congress rep that in this day and age of geolocation, GPS, gee wiz Google knows more about me than my Mother does, maybe, just maybe we could get updated methodologies to calculate the real number of people in this country each month since it's such a critical number upon which so many other ratios are based.
Some ratios we can look at are current rates. For example, the civilian labor participation rate is still horrific.
What about the unemployment rate? Well, the problem here is how many people are really dropping out of the labor force, not to be counted, which is another metric dependent upon overall population.
We'll have more details on the January unemployment report in another post. For now, one can assuredly say the United States did not, as usual, create enough jobs.