The job-filled non-recovery
I used to argue that the whole idea of a jobless recovery was a myth. We weren’t getting many jobs because the recovery wasn’t occurring. Growth was close to trend, so you wouldn’t expect a fall in the unemployment rate.
Now things are beginning to enter the Twilight Zone. Last November the unemployment rate was 9.8%. Now it’s 8.6%, a pretty big drop for one year. Yet according to Okun’s Law the unemployment rate should have risen significantly. Even if you adjust trend growth down from 3.0% to 2.5%, unemployment should have risen, as the four quarter growth in RGDP from 2010:4 to 2011:4 will almost certainly be under 2%.
Explanations? Bad RGDP numbers plus bad unemployment rate numbers plus TGS plus people exiting the workforce. That should just about cover it.
Update: Marcus Nunes has an excellent post with graphs that show the employment picture from a number of angles.
Tags:
2. December 2011 at 10:16
I’m tempted to claim that the unemployment rate is to the labor force participation rate as inflation is to NGDP. That is to say, the wrong object to study and that the differences don’t mater all the time. But when they do, watch out.
2. December 2011 at 10:16
And perhaps Okun’s law never should have been formulated as an unemployment-GDP law in the first place. Perhaps it should have been formulated as an EP-GDP law with the possibility of adjustments for major shifts in preferences (like the rise in female LFP).
The discouraged/unemployed and in LF/out of LF distinction is good enough for government work, but not especially meaningful as a scientific concept – particularly at a time like this.
What does output “cause”? It “causes” employment, not unemployment. If we’re going to have the audacity of talking about “laws” they are probably best put in terms of meaningful scientific relationships.
2. December 2011 at 10:20
Scott,
I put on my Marcus hat and made a graph.
http://img215.imageshack.us/img215/6482/68205248.png
There’s less of a “discrepancy” using this measure of the output gap: the gap is closing, slowly, and unemployment is falling, equally slowly. Perhaps you have to adjust the trend down even more for this recession? A scary thought.
If you extrapolate the 1984-2007 trend, you do indeed get falling unemployment with a widening output gap; a “paradox”.
2. December 2011 at 10:48
Daniel makes a great point. I prefer not to regress changes in the unemployment rate against log changes in GDP growth.
I prefer to think of it as log employment and log GDP being cointegrated. The cointegrating part of the relationship primarily affects employment so that if employment falls, then this will reduce predicted GDP in the cointegrating relationship. If predicted GDP is below actual GDP, then there will be upward pressure on changes in log employment. In other words, if GDP has grown and employment is stagnant, then there will be upward pressure on employment as employers attempt to expand output further.
2. December 2011 at 10:59
A couple of things are happening when people are falling out of the (currently measured) statistics:
1) They may have lost too much money trying to be self-employed.
2) They now know that when they do take a job, someone else might have even needed that job worse than them.
3) The available job, while one is glad to have it, nonetheless after a certain age may threaten them physically in ways that means having to seek out healthcare that would be even more expensive than any wages one could make on the job; ergo it is financially smarter to do work that may not monetarily pay but not pose the same threat in a financial healthcare sense.
4) Or of course the ultimate jackpot: disability (with apologies to those who sincerely deserve it). Voila – better looking statistics. Pardon the snark I couldn’t seem to help myself today.
2. December 2011 at 12:34
I can’t believe you’re talking about RGDP after you banned the i-word! How would you even calculate RGDP without that?
If you can’t put this in terms of NGDP then I don’t really know what you’re saying…
2. December 2011 at 12:49
Brad DeLong emphasizes in a recent post that its mostly driven by labor force participation. The drop in unemployment isn’t quite the good news as I had initially thought. Calculated Risk’s take is likewise sobering. At least holiday shopping is up from last year and we are winning the policy debate.
2. December 2011 at 12:59
Why aren’t the Austrians surprised?
2. December 2011 at 13:28
Maybe … your whole macroeconomic theory is wrong and you should be focusing on PSST type explanations?
j/k, man, you’re good, you can still believe that goosing NGDP is the key to economic health, and that Zimbabwe’s problem was too much overall spending.
2. December 2011 at 13:47
There were something like 150k jobs created and 315k people saying “awww f*** it” and exiting the labor force. Basically there were just enough people getting new jobs to keep up with population growth and the drop in unemployment was almost entirely to entirely driven by people exiting the labor force.
I can’t see this as anything other than terrible news honestly.
2. December 2011 at 14:10
Everyone, I won’t answer comments individually, as I don’t have strong views. I basically agree with everyone except Silas.
Silas, Okun’s Law isn’t part of my macro model.
2. December 2011 at 14:38
Silas, I think PSST can best be seen in the data by dis-aggregating employment and RGDP. For instance, consider the residential investment in RGDP. It is primarily driven by home construction and is a prime driver of construction employment. There’s multivariate co-integration among them all. However, due to the elevated level of inventory, home construction is likely to be weak. This is likely to mean a slow recovery of construction employment, residential investment, and then RGDP growth (presuming other areas of the real economy do not compensate, which is not entirely realistic since business investment should recover more strongly).
Hence, a policy rule that boosts NGDP will have the effect of increasing PGDP, while having a historically weaker-than-expected impact on RGDP growth. As far as I can tell, this is all consistent with the data once you look a little closer at it.
I left out from above that stronger NGDP without a subsequent rise in RGDP will result in higher PGDP, which can also possibly reflate home prices. This would be like a second-round effect that could boost RGDP growth.
2. December 2011 at 16:06
If you use the employment/population ratio instead of the UR, we’ve been essentially flat for two years. This is consistent with NGDP growing at trend, but not closing the wide gap created in 2008-2009.
2. December 2011 at 16:42
@ Integral
Thanks for “honoring” my “hat”. You just developed the “pitchfork theory of unemployment”, a close cousin to my “hole theory of employment”. Nicely done.
2. December 2011 at 17:43
People are beaten. They are going on disability, cash pay, living with relatives etc.
Listen, the regular guys need a prosperous economy to prosper. Not everyone has a Phd. or can slam dunk from 15 feet out. Most people are average–that’s why it is called average. When you get 55, out of work, and have had a desk job most of your life, what do you do?
Inside the hallowed, marbled halls of the Fed, they probably don’t know this.
But I am fine with a few extra points of inflation if it can mean full employment.
2. December 2011 at 20:00
Listen, the regular guys need a prosperous economy to prosper.
Right.
3. December 2011 at 05:02
@ Integral
I just posted a new version of the “hole theory”:
http://thefaintofheart.wordpress.com/2011/12/03/the-hole-theory-of-employment/
3. December 2011 at 05:33
It would be interesting to read a post on which macroeconomic statistics, if any, are reliable enough to draw conclusions from. (You’ve probably written one already?)
3. December 2011 at 06:34
John Hall, Yes, it’s consistent with PSST, but also my money plus structural problems view.
Tommy, I agree.
Marcus, Those are very good graphs, I’ll add a link.
Ben and Catherine, I agree.
Ken, I share your frustration.
3. December 2011 at 06:43
People what are you talking about here ;). Nothing changes.
EmRatio = 58.5
http://research.stlouisfed.org/fred2/series/EMRATIO
The same level as in 2011-03-01, 2010-09-01, 2009-10-01.
http://research.stlouisfed.org/fred2/data/EMRATIO.txt
Nothing changed. NOTHING!
The one thing that changed is a number of discoured people to 64% from 66% in 2008-10-01.
http://research.stlouisfed.org/fred2/series/CIVPART
The real unemloynment rate is still about 16% (discoured + part time).
This indicator is highly misleading.
2011-10: (58,4 – 64,2) / 64,2 = 9.0
2011-11: (58,5 – 64) / 64 = 8.6
Why even talk about it for a second?
Krugman thinks the same:
http://krugman.blogs.nytimes.com/2011/12/02/meh-and-i-say-that-with-feeling/
3. December 2011 at 07:30
Wadolowski, I agree.
3. December 2011 at 13:47
It’s not a job filled recovery just because unemployment numbers drop. Labor force participation is testing all time lows. I think labor force participation is a much better measure than unemployment anyway.
3. December 2011 at 20:50
John, I agree.
12. March 2012 at 16:28
[…] a post I did on December 2nd, and another on February 3rd, on the same […]
31. January 2013 at 06:55
[…] in 2011 I did a post entitled “The job-filled non-recovery.” Now in 2013 The Economist has an article entitled: The Job-Rich […]
23. March 2017 at 05:05
[…] in 2011 I did a post entitled “The job-filled non-recovery.” Now in 2013 The Economist has an article […]