Ugly, just plain ugly -- that's the best way to describe the September employment report. The economy dropped 263,000 jobs in the month, and 7.2 million now since the start of the recession back in December of 2007. The total number of unemployed rose to 15.1 million, an increase of 7.6 million since the recession began. That brought the unemployment rate up to 9.8%.
Silver linings were few and far between in this report. One of the few good news items was that the number of jobs lost in August was revised to 201,000 from 216,000. However, July was revised down to a loss of 304,000 jobs from 276,000.
This is the highest unemployment rate since the middle of 1983. Back in the early 1980’s, demographics (Baby Boomers and women entering the labor force) made the natural rate of unemployment much higher than it is today, so arguably the current situation is worse. It certainly is when measured by the year-over-year change in employment, which fell to -4.23% as shown in the graph below (from http://www.calculatedriskblog.com/). Since 1960, the year-over-year change has never been worse than 3%.
Even worse, those previous valleys had come on the heels of steep mountains of sharp employment growth. This decline is coming on the heels of an expansion that was simply pathetic in the job creation department. Keep in mind that the unemployment rate does not include discouraged workers, or those who are working part-time for economic reasons.
The broader U-6 measure of unemployment rose to 17.0% from 16.8% in August and 11.2% a year ago. Overall, U-6 is probably a better measure of the overall weakness in the labor market than the more widely reported "official" U-3 number.
Average hourly wages crept up by a penny, or less than 0.1%, to $18.67, and are up 2.5% year over year. Remember that those numbers are not adjusted for inflation, and while right now year-over-year headline inflation is low, that is going to change dramatically over the next few months as the plunge in oil prices a year ago slips into the history books.
Actual take-home pay has not done nearly as well though as people are working fewer hours, so average weekly earnings are up just 0.7%. The average work week declined by 0.1 hour to 33.0. While that might not sound like much, remember that there are still 130.9 million non-farm jobs in the country.
That 0.1 hour change in the workweek equates to 395,000 more jobs (33/33.1 x 130.9 million). The workweek data does not go out further than one decimal point, so that 395,000 is just a rough estimate. However, keep in mind that total output is equal to total hours worked times output per hour (a.k.a. productivity), and total hours are made up of the number of people working times the number of hours they work, so the average workweek is very important. It is also an important leading indicator.
When business picks up, most employers will start giving their existing employees more hours rather than go out and hire more new employees. The fact that the average workweek is still declining is a very bad sign.
While the unemployment rate tends to get all the headlines, of more significance to the economy is the employment rate, or the percentage of people who have jobs. Now, that number will never come close to 100%, unless we repeal the child labor laws. Still, one way or another the total population has to be supported by those who are working, either directly as dependents, or indirectly through taxes (Social Security, for example).
There are two related measures. One is the civilian participation rate (red line on the next graph), or the percentage of people who want to work (i.e. are not retired, happy as stay-at-home parents, or still in diapers), and the other is the employment-to-population ratio, or as I like to call it, "the employment rate" (blue line).
The difference between the lines is the unemployment rate. The employment rate is obviously much more volatile than the participation rate, although the participation rate does tend to decline a little bit in economic hard times as people become discouraged, or decide to go to school rather than look for a job. As the graph below shows, both measures were in a secular uptrend until about 2000.
This was driven by the demographics I mentioned earlier: women entering the labor force, and Baby Boomers getting to working age. Now Baby Boomers are on the cusp of retirement, and women are fully integrated into the workforce. Thus the participation rate has started to tail off.
This should make life easier for policy makers, and as it declines, the natural rate of unemployment should come down, all other things being equal. The employment rate never came close to matching its pervious high during the last expansion, and has since fallen off a cliff. It plunged by 0.4 points in September (the graph only goes through August; the St. Louis Fed will probably update the database with the current numbers later today or tomorrow) to 58.8 from 59.2 in August. It is at its lowest point since January of 1984. If that was morning in America, this should be mourning in America.
One of the most frightening things about the current recession (or the one just passed; we are probably technically out of it) has been the rise in the number of long-term unemployed. The median duration of unemployment surged to a record high 17.3 weeks in September from 15.4 weeks in August. A year ago it was 10.3 weeks. The average duration of unemployment jumped to 26.2 weeks from 24.9 weeks in August and 18.7 weeks a year ago. Remember that a year ago we had already been in a recession for longer than either the 1991 or the 2001 recessions, so this is not exactly an easy comp.
Going over 26 weeks for average unemployment is stunning. Keep in mind that regular state unemployment benefits run out after 26 weeks. Benefits have been extended as part of the stimulus package, but even those are scheduled to run out soon for many. The House has approved another 13-week extension for those people in high (over 8.5%) unemployment states, but the Senate has yet to act on the bill. One may think this has to be an urgent priority, as an estimated 1.5 million people are scheduled to run out of even the extended benefits by the end of the year.
What are the odds that those people can continue to pay their mortgages? Not very high, and that will lead to more of them being foreclosed upon, or simply stop paying their mortgages and live as squatters in their own homes until the sheriff shows up at the door. This will lead to more losses at the big banks like Bank of America (BAC - Analyst Report), JP Morgan (JPM - Analyst Report) and Wells Fargo (WFC - Analyst Report), as well as any institution that is holding the mortgage-backed paper. More and more, that is you and me -- through the Federal Reserve, which has been in the process of buying $1.25 Trillion of mortgage backed assets, as well as $200 billion of Fannie and Freddie paper that is indirectly backed by mortgages.
One interesting measure of this increase in the length of unemployment is the ratio between short-term unemployed (less than 5 weeks) and long-term unemployed (over 26 weeks). Prior to this year, the record for that ratio was 0.784 -- hit in March of 1983, and the average since 1960 is 0.369. A year ago, it stood at 0.713. In September, it rose to 1.833 from 1.648 in August. The history of this ratio is shown in the graph below.
In total, 5.44 million people have been out of work for more than 26 weeks, up from 4.99 million in August, and just 2.04 million a year ago. There is some good news in that the number of short-term unemployed is actually down for two months in a row, falling to 2.97 million from 3.03 million in August, and up only slightly from 2.86 million a year ago. This would indicate to me that the pace of layoffs is not as high as it was, but that once unemployed it is becoming increasingly difficult to find a new job.
Right now, it is less a question of high employment destruction than it is an extremely low level of employment creation. In good times and bad both happen, and the employment numbers measure the difference between the two. But the evidence seems to suggest that low employment creation is at the core of the problem right now.
The increase in unemployment was widespread across all demographic groups except Hispanics, who saw their unemployment rate drop 0.3% to 12.7%. Whites saw a 0.1% increase to 9.0% and Blacks saw a 0.3% increase to 15.4%. Teen unemployment rose 0.4 points to 25.9%. Unemployment among adult men rose 0.2 points to 10.3% and among women it was up 0.2 points to 7.8%.
Job losses were also widespread by economic sector, with employment in goods-producing industries falling by 116,000. That was split between a 64,000 decline in construction jobs and a 51,000 decline in manufacturing jobs. Since the start of the recession we have lost 1.5 million construction jobs and 2.1 million manufacturing jobs. We lost 147,000 service sector jobs on the month, including 39,000 in Retail.
The only sector of the economy that was adding jobs was -- surprise, surprise -- Education and Health, and I’ll bet it was mostly health, which added 3,000 jobs. Even the Government is laying off lots of people, mostly at the state and local level. Overall, government employment fell by 53,000 for the month, including 24,000 at the municipal level.
In general, this was a very disappointing report, well below the consensus expectations of a loss of only 180,000, but the disappointment runs much deeper than that -- the internal measures within the report were, if anything, even worse than the depressing headline numbers.
While I still think the economy is technically out of the recession, we are probably entering a long period of a jobless anemic recovery. The graph below (from http://www.calculatedriskblog.com/) shows how the last two recessions were very different from earlier ones, with very slow recoveries in the labor market. Even though both the 1990 and 2001 recessions were very mild in terms of the percentage of overall employment, they were the two longest in terms of the number of months to surpass the previous employment peak (passing back above the 0% line).
This recession is the worst of both worlds. We just surpassed the 1948 (WWII demobilization) downturn in terms of the percentage of total jobs lost (the year-over-year change in the first graph understates things since the decline in employment has been going on now for 21 months). However, the 1948 downturn had just about recovered all of the jobs lost by the 21st month, while we are still falling. The 1990 recession took 30 months to get employment back to where it started. The pathetic recovery after the 2001 recession took 46 months -- almost 4 years to get back to jobs breaking even.
It seems entirely possible that we might not get back to a new record high in total employment until 2015 or so, if this recovery follows a similar path to the previous two. The economic imbalances going into this recession were far more severe than in 1990 or 2001, which would argue for an even slower, more gradual recovery.