The December to January unemployment statistics are often reported wrong in the press. We're sorry, god love ya, but these articles are plain incorrect. People like to compare the month to month change in population, the number of people no longer considered part of the labor force and other data. The grave mistake made by so many in the press and elsewhere is not realizing annual population adjustments* are placed in the January data, not distributed evenly across the entire year, or backwards applied and that's why one cannot compare these two months. Below is a graph of non-institutional population monthly change. This is the number from where all other unemployment statistics are derived. It represents people 16 and older, not locked up somewhere, in a medical facility or in the military.
See those huge three spikes in the above graph? That's when the latest Census, taken every 10 years, has been incorporated into the data series. What happens is almost a do over, starting with the next year and you see a huge discontinuity in the data when the Census has been incorporated into the non-institutional population statistic. Believe me, we did not get a streaming horde of illegal aliens in one month, nor did everyone decide to give spontaneous birth. Those spikes simply represent the tacking on of population controls to reflect the latest Census.
Above is a close up to show the spike in non-institutional civilian population. See how even the monthly change is except between December and January? That's where the yearly population adjustments* are placed in the data series. The adjustments are literally just plopped into the month of January and this is why it is statistically invalid to compare December to January monthly changes. You simply cannot compare a change of a month, when one of those month's includes a year of population adjustments.
To wit, is the current unemployment rate of 8.3% valid? Well, if you like the definitions used by the BLS, it is statistically consistent. Did 1.25 million drop out of the labor force all of a sudden? No, this has everything to do with the Census population levels. Basically there are more people and like many population numbers increasing by proportion with the Census 2010, so did the number of people not in the labor force.
The above is my favorite yearly confusing table from the BLS. This table gives you what the monthly comparisons would be from December to January without the population adjustments. Few read it. Fewer still understand it. Column 1 of table C. is what is reported to you in the press. Column 2 are the population adjustments, put into the January data. The 2012 population controls effect of column 2 is that one lump dump of monthly data which represents a year's worth of adjustments. This time the second column also represents the 2010 Census data incorporated. Ok, now, the last column, column 3. That tells you what the monthly change would be if there were no population adjustments or the 2010 Census was not incorporated into the January numbers.
As you can see, the number of official unemployed would have dropped dramatically, -381,000, the number of employed would have skyrocketed, 631,000 and the number of people who are part of the non-institutional civilian population, but not in the labor force would have dropped by -75,000.
So why isn't the last column used as the official monthly changes? Because this data isn't valid either. You can only go so long without adjusting population growth to actual survey's gathered out in the field without running amok in your numbers and proportions, ratios.
It's true, the not in the labor force monthly increase was a shocker. But if one look at the monthly change, represented by percentages in the graph above, we've actually seen worse. Not in the labor force clearly has people not counted who want a job. It is too statistically noisy to be otherwise.
Here is the BLS explanation on the population adjustments, which so many fundamentally question.
The adjustment increased the estimated size of the civilian noninstitutional population in December by 1,510,000, the civilian labor force by 258,000, employment by 216,000, unemployment by 42,000, and persons not in the labor force by 1,252,000. Although the total unemployment rate was unaffected, the labor force participation rate and the employment-population ratio we re each reduced by 0.3 percentage point. This was because the population increase was primarily among persons 55 and older and, to a lesser degree, persons 16 to 24 years of age. Both these age groups have lower levels of labor force participation than the general population.
Remember, these adjustments represent a decade of change, and the last of the baby boomers are now 50. Yet, we'll be drilling down in the data on this dramatic change, for how many over the age of 55 need to work and are simply being discriminated against?
According to the BLS, of the population adjustment of 1.510 million added to the January non-institutional civilian population, 1.288 million were over the age of 55. Of those, 723,000, or 56%, were not in the labor force. Then, of people ages 16-19, their population adjustment was 430,000 and 327,000, or 76% of them were not in the labor force. To make matters more bizarre, the adjustment to the population assumed to be in their prime earning years, 25-54, dropped by -299,000 in the population adjustments.
Could an additional 1.288 million people over the course of a decade be added to the retirement gang, the 2000 Census off by that many? Yes. Regardless, we'll crank and drill down in those numbers in another post. Stay tuned.
Popular articles which are plain incorrect are listed below.
- ZeroHedge. We're sorry, 1.2 million people did not suddenly drop out of the labor force.
- CNBC, same reason as ZeroHedge.
- Human Events. Gotta watch those population controls.
- Washington Times, ibid.
- Us, unfortunately. No, one cannot claim 1.75 million dropped out of the labor force, especially when one doesn't explain how one pulled that number out of the air.
* The annual population adjustments are based on the last Census results, taken every 10 years, with additional adjustments from birth/death models, migration models are then incorporated to adjust for the total annual benchmark adjustment. To read more about Census Intercensal methodologies, click here.