The November jobs report from the BLS (Bureau of Labor Statistics) came out today, and it included some surprising news. The unemployment rate, listed last month at 9.0%, is down to 8.6%. And yet, the net number of jobs added in the country, according to this report, is about 120,000 - just about enough to keep up with the growth of the labor force. (This replacement number is lower than it used to be, owing to slower immigration reducing the rate of labor force growth). So, what are we to make of these disparate number? Some explanations below.
First, it's important to note that the two figures - the unemployment rate and the number of jobs added - are produced through two different surveys. The jobs number is generated by surveying businesses in various sectors about their hiring and firing activities over the previous month. The unemployment rate number is generated by surveying households, and asking respondents about their employment situation. Surveys are inexact - more on this in a moment - and both of these measures have their own margin of error. If we happen to have a month in which the two surveys err in opposite directions, we can end up with a seeming divergence, like we see this month.
My personal theory is that this divergence gets worse in November and December, owing to temporary holiday jobs. The BLS adjusts its jobs numbers to take seasonal hiring into account. There were a lot more people hired in the retail sector in November than this report indicates in its bottom line number; the BLS does its best to figure how many of those jobs are seasonal, and how many are long-term jobs. This is useful, because the adjustment brings the number more in line with economic trends, canceling out the noise of annual, short-term changes. My pet theory is that The BLS adjustments capture more seasonal jobs in the employer survey than in the household survey. In other words, more the jobs reported by employers are removed from the report than those reported by individuals, resulting in a larger downward revision in the jobs numbers than in the employment numbers.
Secondly, the revisions. When one of these jobs reports comes out, they don't just report their estimate for the previous month's job activity. They also revise the numbers from the last two jobs reports. In this case, they revised the September and October numbers upwards by 70,000 jobs. Remember that horrible August jobs report, which showed 0 net jobs added to the economy? Not so much - that number has now been revised and revised again, to well over 100,000 jobs. (Your tax dollars at work, ladies and gentlemen!)
Similarly, it is likely that this month's report will be adjusted upwards. For instance, the 12,000 net jobs lost in the construction sector just don't match up with the other data about construction trends. Yes, single family home prices continue to swirl around the bowl, but mulifamily housing and commercial construction are seeing significant pickup.
At the same time, the unemployment rate numbers in these reports can be off, too. While these numbers aren't usually retroactively adjusted like the jobs numbers, they tend to get straightened out over time on their own - for instance, one month will show a large drop with just a few jobs added, while the next will show the rate remaining stable even though many jobs were added. We're probably seeing something like this happening in today's job report - a number that reflects not just what happened in November, but which also, belatedly, reflects what happened in the months before November.
It's just my instinct, but I suspect the 8.6% number is a bit optimistic, and that we're likely to see it stay put or rise in the December report, even if there is a great deal of net job creation in December.
What I think we're going to see is that, since July, the economy has added about 700,000 jobs over a five month period. That works out to about 140,000 jobs per month, for five straight months. In July, the unemployment rate was listed at 9.2%
So, to answer my original question, how can 120,000 jobs lower the unemployment rate by 0.4% in a month? They can't. However, 700,000 jobs can lower the unemployment rate by about half a point over five months. We didn't see a big drop in unemployment in November, after four months of stagnation. We saw a slight ticking down in each of the past five months.
The lesson here is that individual jobs reports are just too noisy and imprecise to tell us a whole lot by themselves. It's important to look at trends and aggregates over several months, in order to get a sense of what's happening in the economy.