On February 5th, the U.S. Bureau of Labor Statistics (BLS) reported that the unemployment rate fell to 9.7% from 10.0%. While I am more than happy that fewer people are unemployed, I have reasons to be skeptical. The problem is that we read news reports of continued job losses, yet we also hear that the unemployment rate is falling. It seems that something does not add up. Let’s see if we can take the massive amount of data that is available at the BLS, perform some simple visualizations, and see if we can understand why continued job losses and lower unemployment can happen at the same time.
We recently reported on comparing job losses with the President of the United States. While the chart has a lot of spin in it, the one interesting trend that becomes apparent is that while job losses are continuing, they are not continuing at the same frantic pace as a year ago. This means things are getting worse at a less rapid pace. Said another way, job losses seem to be stabilizing. That is good news indeed.
However, we recently reported that 824,000 jobs were going to be lost. Also, the BLS reported that 20,000 people lost their jobs last month. Additionally, the BLS revised the job loss numbers that it reported for last year. As you can see from the bar graph below, the number of jobs lost last year increased in every month except for November 2009. In November, the BLS report 64,000 jobs gained versus a previous 4,000 jobs gained. This was probably due to temporary hires for the Christmas season.
Finally, former Federal Reserve Chairman Alan Greenspan has said: “But remember that unless there is a monthly increase of more than 100,000 a month, you’ve still got the unemployment rate continuing to rise.”
With all of this, we still need to answer the question: how can jobs be lost and the employment rate fall?
The reason that the employment rate fell while job losses continued is that the Current Employment Statistics (CES) is where the jobs number comes from. This is a survey of approximately 400,000 businesses. It is this survey that is showing all the job losses. However, this survey has nothing to do with the unemployment rate.
The unemployment rate is calculated by the Current Population Survey (CPS), which is commonly called the Household Survey. The unemployment rate is not calculated by any of the revisions mentioned above. The BLS surveys approximately 60,000 households every month, and asks about the employment status of every person 16 years and older. Based on the responses given, a person is classified as either unemployed, employed, or not in the labor force.
One survey, the CES, can easily say that we have more job losses while another survey, the CPS, can say that we have lower unemployment.
The unemployment rate is the number of people looking for job, but not yet employed, divided by the civilian labor force. If a person becomes discouraged and stops looking for a job, then they are no longer counted. Thus the unemployment rate is the number of unemployed divided by the civilian labor force. The civilian labor force held relatively stable at 153.17 million people (it has varied by plus or minus one percent over the last two years). If the denominator stayed roughly the same, then to cause the drop in unemployment, the number of unemployed must have declined.
But this still leaves an open question. Why did unemployment fall in the CPS survey? There are some possible answers. One possible answer would be that the unemployed did not look for work in the last 4 weeks. That would disqualify them from being unemployed. Another possible answer would be that the unemployed were looking for a full-time job, but found a part-time job instead. That would disqualify them from being unemployed.
Yet another possibility is that the Seasonal Adjustments to the unemployment rate are hiding the true picture. If you take a look at the following graph, we can see a plot of the Seasonal Adjusted Unemployment Rate versus the Non-Seasonal Adjusted Unemployment Rate. As workers are hired for the Christmas season, the unemployment rate temporarily drops. Every January, there is a spike in the unemployment rate as temporary workers are laid off after the Christmas season. For example, the Non-Seasonal Adjusted Unemployment Rate in January 2010 was 10.6% versus the Seasonal Adjusted Unemployment Rate of 9.7%
To smooth out the bumps caused by this, the BLS seasonally adjusts the numbers. There is nothing wrong with these adjustments, as long as the BLS is consistent with them.
Looking at the chart, you can see that the seasonal adjustments are increasing in amplitude. Now here comes the million billion trillion dollar question. Since the leg up on the seasonal adjustments is so large, will the downward leg be just as large? In other words, will the seasonal adjustments make the unemployment rate worse in the coming months? We will have to wait and see.
The idea for this post came from Mish and his two posts here and here. But I did not want to take his word for it, I wanted to take a look at the data myself. Fortunately the BLS has it all on-line. Oh, and the play on words in the title, it comes from here.
