The economy added a total of 36,000 jobs in January. That is far below consensus expectations for a gain of 148,000. The small increase is particularly disappointing relative to the numbers put out on Wednesday by ADP that showed a gain of 187,000 private sector jobs.
Government payrolls declined by 14,000 and the private sector added a total of 50,000. The consensus was looking for a decline of 15,000 Government jobs and thus also for a gain of 163,000 private sector jobs.
On the other hand, the unemployment rate, which is derived from a separate survey showed a surprising drop to 9.0% from 9.4% in December, and 9.8% in November. That is the lowest rate since April 2009. That is also the sharpest two month drop in the unemployment rate since November 1958 (the month I was born!).
Earlier Months Revised
The numbers for November and December were revised higher. In November we actually gained 93,000 jobs, not the 71,000 reported last month. This was the second upward revision of the November number, it was originally reported at just 39,000 jobs added. In December we gained 121,000 jobs, not 103,000 as was reported last month. Almost all of the upward revisions came from the private sector.
The private sector actually added 128,000 in November, not the 79,000 reported last month, and far above the 50,000 that was first reported. December the gain was 139,000, not 113,000. These upward revisions take a bit of the edge off or what is at first blush a very disappointing number from the establishment survey. But just a bit, as even if you add the upward revisions to the January total, you only get a total gain of 77,000 jobs, which is still far below what was expected.
Unemployment Rate
The unemployment rate is derived from a separate (household) survey from the total number of jobs (establishment) survey. The household survey actually pointed to a gain of 117,000 of jobs in January, on top of a gain of 297,000 jobs in December. The number of people unemployed according to the household survey fell by 622,000 in January after a decline of 556,000 in December.
However, generally the numbers from the household survey are considered less reliable than are the numbers from the establishment survey. That does not mean they should be disregarded entirely, and the divergences between the two series are often the biggest near turning points in the economy. The household survey does a much better job of picking up people who are self employed, and of very small start up businesses than does the establishment survey.
The unemployment rate fell to 9.0%, it was 9.8% in November. A year ago the unemployment rate was 9.7%. While that is good news, there is less to it than meets the eye. The civilian participation rate, or the percentage of people in the labor force, both employed and unemployed fell to 64.2% in January from 64.3% in December and 64.5% in both November and October.
A year ago it was 64.8%. It is not all a mirage though, just some of it. The employment to population ratio, or the employment rate, actually rose to 58.4% from 58.3% in December and 58.2% in November. It was 58.2% a year ago.
Average Work Week
For all employees,the length of the average work week fell to 34.2 hours from 34.3 hours in November, matching the October level. It is up from 34.0 hours a year ago For production and non-supervisory employees, the length of the average workweek fell to 33.4 hours from 33.5 hours in both December and November. A year ago it was at 33.3 hours.
While an increase of 6 minutes a week might not seem all that significant, but it really is when you multiply it times the 130.265 million people that were working in October (per establishment survey). Average hourly earnings for all employees rose $0.08 to $22.86 on the month and are up 1.9% from $22.44 a year ago. Average hourly earnings for production employees rose $0.10 to $19.34 on the month, and up 2.3% from $18.91 a year ago.
The year over year changes are not great, but then again, inflation is pretty low as well. It is not inflationary from a cost push point of view because it is less than the rate of productivity growth (see here). The monthly increase is a big acceleration from the trend, particularly for hourly workers. If annualized the increase would work out to a rate of 6.4%, and for all employees the annualized rate would be 4.3%.
Annualizing monthly changes is a bit of a dangerous game, so look at those numbers as suggestive, not definitive or predictive. If it can be maintained (a big if) it would be very good news indeed. Income growth in the middle and lower half of the income distribution has been sorely lacking, not just recently, but for over a decade. Higher incomes for those who are working means higher sales and more quickly repaired household balance sheets. That should translate into higher demand, which in turn should help job creation.
While the unemployment rate is better than a year ago, some (not all) of that is a mirage due to falling participation rates. A falling participation rate is not exactly a new development, it has been in a downtrend for a decade now, but the decline has been very steep in the Great Recession and has yet to really turn around.
Employment Rate Rises, Participation Rate Falls
While the unemployment rate gets the headlines, it is worth digging just a little bit deeper into the number. The unemployment rate is really the civilian participation rate divided by the employment rate, also known as the employment population ratio. The total population is divided into three groups, the employed, the unemployed and those not in the workforce.
The participation rate (blue line in the graph below) is the percentage that are either employed or unemployed. It will never reach 100%. For that to happen, we would have to do away with all child labor laws and insist that those lazy 2 year olds stop napping and get to work. The Social Security retirement age would have to be raised not to 66 or 67, but to 166. The highest the participation rate ever reached was 67.3% in April of 2000.
The participation rate will normally slump during a recession and its aftermath. However as the first graph below shows, the participation rate was in a huge secular increase from the mid 1960’s until the end of the 20th century. Yes it would flatten out and decline slightly during recessions, but it would always return to a higher high, and the low during the next recession was always much higher than the previous low. That did not happen in the last expansion.
The highest the participation rate hit during the last expansion was 65.8% in January 2005. In January, the participation rate fell to 64.2% from 64.3% in December (and 64.5% in November) and 64.8% a year ago. The employment rate though, increased for the second month in a row to 58.4% from 58.3% in December and 58.2% in November, but remains below the 58.5% rate of a year ago.
The Historical Context
The secular rise in the participation rate was due to two huge demographic trends. First was the entry of the baby boomers into the workforce. Remember you are neither employed not unemployed when you are a kid. The baby boom started in 1946, so by the mid-1960’s they were reaching the age when they were either employed or unemployed (or out of the country in Vietnam). That was a major force lifting the participation rate until the early 1980’s.
The second major demographic force that started just a bit later (in force) but continued longer was the increased participation of women in the labor force. Back in the mid 1960’s if a magazine article mentioned the words woman, and labor in the same paragraph, the odds were that the article was about childbirth. That clearly is no longer the case today. In January, there were 63.4 million women working (over age 20), not that much behind the 71.6 million men with jobs (per household survey).
The front end of the baby boom is just now hitting retirement age and that will put continuing downward secular pressure on the participation rate for years to come. The participation rate took a big dive during the early part of the recession, started to rebound earlier this year, but then started to drift lower in June and July. It ticked up in August, and managed to hold onto that increase in September, but fell again in October. This month it fell to 64.2%.
A rising participation rate will put upward pressure on the unemployment rate, but should nevertheless be considered to be good news. A falling participation rate will lower the unemployment rate, but is bad news for the economy.
The other side of the decomposition of the unemployment rate is the employment to population ratio, or the employment rate (black line). That is the percentage of the population that actually has a job. One way or another, these are the people that have to support the rest of the population.
This is a hugely under reported number, and one that deserves a lot more attention than it gets. Like the participation rate, it had been in a secular upward trend from the mid 1960’s through the end of the century. It is however much more volatile than the participation rate (it has to be, if it always moved in tandem with the participation rate, the unemployment rate (red line, right hand scale) would never change). Its’ high water mark was 64.7% in April 2000.
Unlike previous recoveries, it never came close to hitting a new high after the 2001 recession was over, only getting back to 63.7% in March of 2007 before starting to fall again. During the Great Recession it really fell off a cliff, hitting 58.2% a year ago. It has erratically pushed its way higher so far this year and hit 58.5% in August, and remained there in September.
Unfortunately the decline resumed again in October, falling to 58.3% in November it fell back to 58.2%. That was the lowest percentage of the population working since November of 1983. It has however ticked higher in each of the last two months, and is at 58.4%. That is still a lousy level, but at least it is moving in the right direction, and indicates that at least part of the drop in the unemployment rate is for real. Graph from http://www.calculatedriskblog.com/.
Better Than the Last Two Times
Note that in the 1991 and 2001 recessions, the employment rate continued to decline for a very long time after the recession ended. The NBER declared the Great Recession officially over as of June 2009. You would never know it from listening to the press or the pundits, but this recovery has been significantly better on the jobs front than the two recessions that preceded it, particularly when it comes to private sector employment.
To my mind the employment rate should get more attention than it does, and the unemployment rate less, although clearly the two numbers are related. The unemployment rate though can be more subject to distortions than can be the employment rate. That was proved true in November when it rose sharply, and is also true this month when it plunged.
As a matter of economic history, it should be noted that both Presidents Carter and Reagan get a bit of a bum rap when it comes to the unemployment rate. When the participation rate is rising, the economy has to produce significantly more jobs to keep the unemployment rate from rising. On the other hand, the second president Bush gets way too much of a free ride when it comes to the unemployment rate, since the participation rate was falling for most of his time in office. The same is true for Obama.
Duration Measures Deteriorate
There was mixed news on the duration of unemployment front. Over time, the number of short term unemployed really does not vary that much. People are always losing job, or in boom times, quitting jobs. Next week we should get the Job Openings and Labor Turnover Survey (JOLTS) which will tell how many people are getting laid off versus quitting and the actual number of new jobs created. The numbers today simply show the net difference between jobs lost and jobs gained, rather than the totals for each side. Unfortunately, the JOLTS data will be for December, not January.
It is the number of long term unemployed that really make the difference between boom and bust. The extraordinarily long time that people have been out of work after they lose their jobs is what has really set this recession apart from all the previous pre-war recessions. Over the summer we had a couple of months of good news on that front, after two years of absolutely horrifying numbers. This month brings more bad news in this regard. .
The average duration of unemployment (red line) rose to 36.9 weeks from 34.2 weeks in December. That is nominally a new record, but the definition was changed this month, so the historical comparison for the average is now meaningless, or at least has to be taken with a bag of rock salt. Previously, if someone reported being out of work for more than 2 years, the BLS would enter them in the database for being out of work for 2 years. This month the maximum was changed to five years.
While an interesting statistic, and relevant in the face of the “99ers,” or those who have been out of work so long that they have exhausted even their extended unemployment benefits, it means that you should not take the increase from last month at anything close to face value. Still, for those who are interested, we are well above the 30.5 week level a year ago, and at the time, that was an all time record. Prior to the Great Recession the previous all time record high was set in June of 1983 at 20.8 weeks.
The median (blue line, half above, half below) duration will always be lower than the average duration since it is impossible to be unemployed for less than zero weeks. It is also not distorted by the change in “top coding” of the data. Its history is not quite as long as the average. There the news is a bit more encouraging. It fell to 21.8 weeks, its first decline after four months in a row of increases. It was 22.4 weeks in December and 21.7 weeks in November and an interim low of 19.9 weeks in August. It is still well below the 25.5 weeks (all-time record) in June. It is still higher than it was a year ago when it was at 20.0 weeks. Given the change in the definition for the average, the median is the more reliable statistic right now.
Prior to this downturn, the highest the median duration had ever hit was 12.3 weeks in May of 1983. Note that it is normally the case that the duration of unemployment continues to rise even after the recession ends. This happened not just in the last two recoveries, but in all post war recoveries. However, following the 1991 and 2001 downturns the persistency of high and rising unemployment duration was much more pronounced than in the earlier downturns. This time, while the peak was an Everest relative to any previous experience (except perhaps for the Great Depression, but the data is not available).
Long term unemployment is a very different experience than short term unemployment. It is not just an unplanned vacation, it is an existential threat to your standard of living. When you lose your job you don’t know how long it will take you to find a new one. You get unemployment insurance benefits (usually, but not always) but in general, they cover just 60% of what you were earning when you were employed, up to a cap of around $400 per week (varies a bit by state). The nationwide average is about $300 per week. Thus for most, the pay cut is much more than 40%. Most people have fixed, or at least semi-fixed expenses that use up more than 60% of their income. They thus have to dig into their savings and/or run up their credit cards. It is also much harder to get a job if you have been out of work for a year than if you have been out of work for just a month or so.
Regular state unemployment benefits run out after 26 weeks, and after that people move over to extended benefits which are paid for by the Federal Government, and which this time around have become a political football. By the point that people get to the six month mark of joblessness, they have usually depleted most of their savings outside of their 401-k or IRA plans, and may well have started to dip into those as well (in the process paying a 10% penalty plus having the withdrawals taxed as ordinary income). That is particularly true this time around because going into this recession the savings rate was at a historic low.
In past downturns, the unemployed who were also homeowners could generally tap into their home equity to tide them over. With 25% of all homes with mortgages now underwater, and another 5% with less than 5% positive equity, that option is no longer available for millions. Thus, without extended benefits these people would be left with no financial resources at all. However, the extended unemployment benefits are divided up into several different tiers. Even though benefits were extended for another year, the maximum is still at 99 weeks. The extension was more about people who have been out of work for 40 weeks or 60 weeks continuing to get benefits, not people getting benefits beyond 99 weeks.
The census bureau tracks four different groups by length of unemployment. The short term unemployed are those that have been out of work for less than five weeks (blue line). Almost always this is the largest group of the unemployed. The next biggest group is usually those that have been out of work between five and 14 weeks (red line). Being out of work for a month is really not that big a deal, but as the joblessness stretches on it becomes a bigger and bigger problem. Not only do your finances start to run dry, but your contacts start to dry up and your skills start to wither. The longer you are out of work, the lower your likely salary once you return to work. Normally the next two groups, those out of work for 15 to 26 weeks (green line), and those out of work for more than 27 (orange line) weeks are a very small proportion of the total unemployed.
That changed in a very big way during this downturn, and in January, 6.210 million, or 43.8% of the 13,863 million total unemployed have been looking for more than 26 weeks. That is a decrease of 231,000 people from last month, well off the peak in May, when there were 6.763 million very long term unemployed. A year ago there were 6.302 million people, or 41.5% of the total unemployed, that were out of work for more than 26 weeks.
The numbers in the graph below are not adjusted for population growth, so we should expect to see a bit of an upward tilt in all four groups over time. Sill, in a healthy economy, the number of very long term unemployed should be down closer to 1 million, not above 6 million. On the other hand, the 2.678 million who are out of work for less than five weeks is actually lower in absolute terms than the average of the last 40 years (despite population growth over that time). Note that in every prior post war recession, the number of short term unemployed remained the largest group. The massive number of long term unemployed is really what sets this downturn apart from all the previous ones.
Overall, this was a very confusing report. It almost seems as if the two surveys, the household and establishment, were looking at two entirely different countries. The establishment numbers were disappointing, even with the upward revisions to the prior two months. Bad weather might have been a factor, but since these are seasonally adjusted numbers, the weather effect was probably not that large. There does seem to be a recent pattern of low initial reports in the establishment data and subsequent upward revisions.
The household survey data, on the other hand, was extremely positive, and not just the fact that the unemployment rate fell to 9.0% when the consensus was looking for it to tick up to 9.5%. This is the second month in a row of a big divergence between the two data sets. Yes the participation rate does continue to fall, but we now have two months in a row where the employment rate has actually increased.
I will have a second post out shortly where I will dig deeper into the numbers and look at the demographics of joblessness as well as where the jobs were being created and lost by economic sector.
Zacks Investment Research
