The job-filled non-recovery continues
A few months back I noticed an odd disconnect between the GDP and employment data. Not only are we not having a “jobless recovery”, we seem to be having something close to the reverse. Obviously I was being a bit provocative; the job growth isn’t all that impressive. But it certainly is impressive when compared to the non-growth in GDP.
Two months back I did a comparison of the unemployment rate and GDP growth since QE2 was announced on November 2010. One year later unemployment had fallen by 1.1%, from 9.8% to 8.7%. The just announced January numbers show unemployment falling to 8.3%, a surprisingly steep fall from 14 months ago. Why do I call it “surprising,” after all, isn’t unemployment supposed to fall during recoveries? Yes, but we haven’t had any recovery over the last 14 months, indeed the economy seems to be slipping ever deeper into recession. The RGDP growth rate has been 1.6% in the 4 quarters since late 2010, which is below everyone’s estimate of trend growth (except perhaps Tyler Cowen, who should be ecstatic about the jobs/GDP disconnect—as it will help him sell more copies of The Great Stagnation.)
When really surprising data appears, the most logical explanation is probably a little of everything. If a single factor could explain this huge anomaly, then you’d expect huge anomalies to be much more common. So what are the likely explanations?
1. Trend growth really is slowing somewhat.
2. People are leaving the labor force.
3. RGDP data is measuring “payroll GDP,” not household GDP
So let’s look a bit closer at the various employment data:
In the 14 months before and after QE2:
Variable 9/09 to 11/10 11/10 to 1/12
Unempl. rate unchanged fell 1.5%
Employment +162,000 +2,700,000
Nonfarm payroll +492,000 +2,183,000
Labor Force +255,000 +354,000
Unemployed +92,000 -2,346,000
RGDP growth 3.3% annual rate about 1.6% annual rate (incomplete data)
A few comments.
The drop in unemployment isn’t just people leaving the labor force, there are lots more people employed. But part of it is the labor force, which is growing less than it should. How much less is difficult to say, because we honestly don’t know how the labor force would have evolved had the recession not occurred.
It’s also not clear whether to look at payroll data or the household survey. Everyone seems to agree that the payroll numbers are better for month-to-month changes, as the sample is more comprehensive. But that’s less clear for changes occurring over 14 months, when sampling errors tend to even out. Remember that the household survey includes people working at small new businesses and the self-employed, which are not picked up in the payroll survey. Perhaps the RGDP data is just picking up “payroll GDP”, not “household GDP.” But even that wouldn’t explain all of the difference between jobs and RGDP.
One thing is clear; the level of employment over the past 14 months has grown well above trend. And real GDP has grown well below trend. Two months ago I thought that was a quirk in the data that would soon reverse. Instead the gap has gotten much bigger. My hunch is that all three of the factors I cited play a role in this anomaly.
Does this tell us anything more about the effectiveness of QE2? I don’t think so, as the data is ambiguous. In any case I’d look at NGDP growth, which has been slow since late 2010.
Tags:
3. February 2012 at 08:11
What is unemployment rate assuming no one dropped out?
3. February 2012 at 08:23
I think this is mostly just give back for the opposite phenomeon awhile back, where working hours were being slashed despite output growth of 2-3% in 2009.
From 09Q2-10Q1, nonfarm business productivity grew at an average quarterly pace of 6.15%, output by 2.78% and hours worked -3.15%.
From 09Q3-11Q4, nonfarm business productivity grew at an average quarterly pace of 2.4%, output at 3.32%, and hours worked at 0.94%.
These won’t sum precisely because they are just simple averages of rounded quarterly quarterly growth numbers, but to me it seems that firms cut themselves to the bone in ’09 and are restaffing now. Arnold Kling for the past several years argued that today, more and more businesses here workers to provide ‘organizational capital’, citing Garrett Jones. They can be more flexible with these types of workers than production workers (you can fire them without sacrificing output, and you can hire them for reasons other than increasing output), and so employment won’t be as closely tied to output in any given period as it was in the past.
http://econlog.econlib.org/archives/2010/03/forecast_for_a.html
The employment to population ratio was 58.5% in Oct09, it was 58.5% in Aug10, and it is still 58.5% as of Jan12. From 2009Q4-2011Q4, GDP growth has averaged about 2.5%. Given that employment has kept up with population growth, that’s consistent with 2.5% being trend GDP growth.
3. February 2012 at 08:31
-“What is unemployment rate assuming no one dropped out?”-
Depends on your assumptions. If we assume that the labor force as of June2008 wasn’t much impacted yet by the recession, and that it would grow by 70k/mo (which takes into account Boomer retirements, maybe aggressively so), it should have hit 157.326 million in Jan12, whereas it was actually 154.359 million, a 2.931 million gap. Add that value to unemployment (12.758 million, so 15.689 million with that extra portion), and divide by the higher labor force of 157.326 million, and we get an unemployment rate of 10.0%. I’d put a plausible range of 9.2%-11.3%, based on a range of 40k/mo labor force growth and 125k/mo.
3. February 2012 at 08:36
We would see a sustained secular bull market if Bernanke would take off the diapers and short pants and get serious about growth.
The econo-shamans are chanting verse about invisible inflation, and theo-monetarists are genuflecting to gold or paper currency—but that is not what a central banker should do.
This is a weak-turd recovery, while the Fed mysteriously frets and ninnies about inflation. The braying loon Plosser is ever whimpering and sniveling about inflation—despite the fact the CPI has been flat to down for three months.
3. February 2012 at 08:48
Can lower unemployment numbers be explained by sticky wages? Wages for the employed are rising at a rate lower than NGDP. The difference between the two is being used to hire more people.
3. February 2012 at 08:53
http://www.zerohedge.com/news/record-12-million-people-fall-out-labor-force-one-month-labor-force-participation-rate-tumbles-
Thanks Justin,
“I’d put a plausible range of 9.2%-11.3%, based on a range of 40k/mo labor force growth and 125k/mo.”
Scott, that basically explains it – doesn’t it?
3. February 2012 at 09:10
What if it was German style “internal adjustment” through falling unit labor costs … (a bit like what Scott N said). The economy readjusting through the price system, even in wages, and without the help of inflation – I know, it’s the thing that’s not supposed to be possible…
3. February 2012 at 09:11
Scott
Tyler Cowen goes is “ecstatic” in a different way:
“All good news, 243k up but lots more information in the numbers, try @JustinWolfers or @BetseyStevenson for details and interpretation. The “big loser” here?: Old Keynesianism. You really can get a recovery when the real shocks are moderately positive. You will note, as we have been told many many times by many many sources, fiscal and monetary policy have not been extremely pro-active in recent times; in fact the stimulus has been trickling to a close. The big winners, apart from the American public?: real business cycle theory. It is part of any cyclical explanation, whether one likes it or not”.
Another big loser is those liquidity trap theories which tell us that positive real shocks are bad for the economy because the AD curve has a perverse slope, etc., and that negative shocks might help spur recovery. That theory is looking very weak, again. I consider it the weakest economic theory that has any currency in the serious economics blogosphere.
3. February 2012 at 10:02
Can’t the liquidity trap people (but also Scott) point to this and this, saying that the gap between the (negative) interest rate we need (in an interest rate targeting only policy) and the zero interest rate bound has decreased?
3. February 2012 at 10:39
Clowns to the left of me, jokers to the right…
Keynesians would call it a jobless recovery if GDP rose while employment didn’t.
Pseudo-Monetarists would call it a job-filled non-recovery if NGDP didn’t rise while employment did.
Why don’t we just settle this once and for all and have Keynesians (pseudo-Monetarists) admit to having to say that they believe a recovery would be had if GDP (NGDP) rose even though the entire world’s population was wiped out except for just two people, one of whom spends $14 trillion on the other, and then vice versa, each time printing and spending 5% more than the last period?
At least it would make it obvious that recovery to them is nothing to do with people, production, output, you know, the stuff of economics, but rather some singular aggregate statistic fetishism akin to mystical worship.
3. February 2012 at 10:58
Benjy said:
“We would see a sustained secular bull market if Bernanke would take off the diapers and short pants and get serious about growth.”
Are you suggesting this bull(shit) market rally can’t be sustained?
Sorry fella. You won’t see any real, sustained growth in the economy until we let banks fail and clear gargantuan debt loads everywhere, no matter what Bernanke does.
And for the rest of you debating employment figures, you may want to consider whether these numbers are of the massaged variety or not. As you may have guessed, I think they are one of the few things being manufactured in this country right now.
3. February 2012 at 11:12
Are expectations adjusting to the new NGDP trendline?
3. February 2012 at 11:21
We are constrained on the oil supply. US oil consumption was off 1.6%, vehicle miles traveled (VMT) down 1.3% in 2011 compared to 2010. (Normal VMT trend would have been +2.4% in 2011, so the delta to trend is 3.7%. That’s a lot!)
Does oil matter? Does mobility matter? My admittedly crude back-of-the-envelope calculation suggests oil tamped down GDP growth by 1.5-1.7% in 2010.
What’s your number?
3. February 2012 at 11:43
Demographics.
3. February 2012 at 12:10
Morgan, See Justin’s answer.
Justin, Interesting comments, although those time periods don’t quite correspond to mine.
Ben, I agree that monetary policy has held us back. The recession began in December 2007. It should be over by now, it’s 2012 already.
ScottN, Yes and no. I’d say that wages were above equilibrium after NGDP plunged, and that’s why higher NGDP doesn’t mean higher wages right now. The equilibrium wage is still below the actual wage.
Morgan, No, you’ve had 2.7 million new jobs in 14 months and 1.6% RGDP growth. Those numbers are totally out of step, and discouraged workers has no bearing on that anomaly.
mbk, Yes, but that’s not the question. I’m not trying to explain why we have so many new jobs (that’s easy to do) but why we have all those new jobs and keep sliding deeper into recession (in terms of GDP.)
Marcus, I think he went overboard. New Keynesian theory predicts the economy will recover on it’s own, even with no stimulus. The mystery isn’t the fast recovery, it’s the slow recovery. Only 1.6% RGDP growth over the past 12 months.
Perhaps he’s focusing on Krugman, who’s bearishness might have been a bit overdone.
John Thacker, See my previous two answers.
Major Freedom, You said;
“Pseudo-Monetarists would call it a job-filled non-recovery if NGDP didn’t rise while employment did.”
You still don’t understand the difference between RGDP and NGDP?
Neal, Steven, and Robert, See my earlier answers.
Everyone, I’m not trying to explain growth, I’m trying to establish whether there has been any growth.
3. February 2012 at 13:23
“Pseudo-Monetarists would call it a job-filled non-recovery if NGDP didn’t rise while employment did.”
“You still don’t understand the difference between RGDP and NGDP?”
You’re right, I am completely clueless. Totally unaware. Real versus nominal? What’s that? Don’t we eat money?
I used to think one was real spending and the other was nominal spending, but then someone, I won’t say who, said that a fall in consumer spending and rise in their cash balances somehow corresponds to no change in aggregate demand, on the very weird basis that consumer goods inventory builds up; as if the mere physical presence of inventory that accumulates on account of a drop in consumer spending is supposed to show that there is really no drop in aggregate demand and thus no drop in spending. Consumers are building up their cash balances and spending less, and yet, as if by magic, aggregate demand goes unchanged because of some reference to physical goods going unsold. That means real goods (inventory) are the same thing as demand (spending on real goods).
Heck, why not lump in accumulating unemployed workers while we’re at it? We can say even though employers dropped their spending on labor and accumulated cash balances, there was a corresponding physical build up of unemployed workers, and so that means aggregate demand is unchanged. S=L!
Then there are the claims that boosting nominal spending necessarily boosts real productivity, and that declines in NGDP “explain” real output recessions, rather than being just another effect of the same thing that causes real output recessions.
On this blog, real and nominal are basically interchangeable. Why can’t others join the insanity too?
3. February 2012 at 13:23
Why did we have growth in 2010 and job creation in 2011? I’d just argue employment lags most everything else. Real growth in 2010 is going to generate a few jobs in 2011. And three years of nominal growth is slowly shrinking the gap between sticky market wages and equilibrium wages.
Why do you we have slow real growth? I’m not too sure we do. I just went to FRED and RGDP tanked around 2011Q1, but it trended up the rest of the year. Tyler is right to some extent and I think 2-2.5% real growth will the new long term trend. But that dip in late 2010/early 2011 seems like an anomaly.
3. February 2012 at 13:44
Variable 9/09 to 11/10 11/10 to 1/12
Unempl. rate unchanged fell 1.5%
RGDP growth 3.3% annual rate about 1.6% annual rate
In the first period, people got a 2% reduction in the payroll tax, which maps to about a 3.3% increase in take home pay for people who already have jobs.
In the second period, unemployment fell by 1.5%, which translates to a 1.6% increase (91.4/90.0) in the percentage of people who want to work who actually find work.
Gee, 3.3% and 1.6% are the RGDP numbers! Coincidence, yes. Creating confidence for the already employed is a bigger stimulus in the short run, but actually employing people is better for the long run. Also the NGDP numbers are being deflated by gas prices to produce RGDP. The Fed has nearly zero ability to influence gas prices despite all the protestations to the contrary.
3. February 2012 at 16:20
Scott,
Might the record setting deportations have anything to do with this phenomenon?
3. February 2012 at 17:42
I am pondering the use of the word “anomaly” in this post. Perhaps there is a more nuanced meaning to it or perhaps I don’t really understand? I have been getting more email from headhunters lately, not much more, one or two a month when I used to get several per week prior to the recession. Over the last couple of years I thought I dropped off their lists because I didn’t get any. So at least from my perspective something about the improving employment numbers is real, I just don’t know how much. I suppose when I land a better gig I won’t have any doubt.
3. February 2012 at 18:44
B, RGDP is up only 1.6% in the last year, so it’s not just the first quarter. But I agree with you the RGDP numbers may be wrong.
Oddly, Tyler Cowen is more likely to be right about the Great Stagnation if he is wrong about the robust recovery.
Steve, I’m afraid you lost me.
Benny, If might explain the labor force decline (or slow growth) but not the weird mismatch between jobs and RGDP.
Bonnie, Good luck.
4. February 2012 at 05:18
I’m not entirely sure on the employment issue, as it’s actual increase would have given a rise to the velocity of money, which it didn’t: http://im-an-economist.blogspot.com/2012/02/graphs-of-week.html
In fact, velocity is still decreasing. This could perhaps mean that even though the people got the jobs, they’re still reluctant to spend their money. We could therefore give them some time to restore their confidence and start spending, which would imply an increase in aggregate demand and consequently the RGDP.
On the other hand, according to what the leading indicators say, the recovery isn’t due for quite some time: http://im-an-economist.blogspot.com/2012/02/business-cycle-tracking-usa.html
4. February 2012 at 07:58
as a few pointed out, productivity during the early stages of a recovery tends to be stronger, so some of the employment growth was catch-up as businesses felt the recovery was more durable (some of the early-stage productivity gains are not sustainable e.g. overtime).
Since the 3rd quarter of 08, the “okun’s law” relationship has been close to average (which itself is pretty noisy quarter to quarter). Also, there is a better relationship between ngdp and employment growth (establishment survey) than rgdp and employment growth. looks like a lot of catch-up employment growth happened in 2nd quarter of last year too.
4. February 2012 at 08:37
[…] […]
4. February 2012 at 17:42
-“Justin, Interesting comments, although those time periods don’t quite correspond to mine.”–
Sure, I was trying to highlight in particular the disconnect between business output growth early in the recovery (+2.8%) and labor hours (-3.2%). Your first time period added many months of job creation during 2010, as well as quarters where GDP grew at a slower pace, which obscured the disconnect.
Real output growth of +2.8% in the business sector is probably around trend. Seeing employment collapse during four quarters of trend growth should be just as puzzling as above trend employment expansion during positive but below trend growth.
I can’t be sure of what exactly happened. Could be that we have more of a Garrett Jones economy, and so employment isn’t as closely tied to production as it was in the past. Businesses might have just been scared to death about the economy in 2009 and were collectively able to overload existing workers, for a time, before gaining confidence in the economy and adding new employees. Maybe many new innovative ideas forced by the crash really did permanently boost worker productivity by 6% from 2009Q2-2010Q1, and the idea stream is tapped out, for awhile, so even mild demand growth required a bunch of new workers. Could be something else. Could be a mix.
I guess my point is that both of your periods taken as a whole seem very reasonable to me – 2.45% GDP growth, and 102,000 new jobs per month. Feels like growth of each was around trend rates.
4. February 2012 at 18:40
Justin, I agree. Taking the two periods as a whole, 2.45%/year RGDP growth and 102,00 jobs/month seems right on track with the Bernanke recovery we’ve come to expect.
Growth was also back on track in 2011Q4. Let’s not read too much into a few transitory trends of the past 10 quarters. Let’s turn our focus back on NGDP and not get spooked by hiccups in RGDP and employment.
5. February 2012 at 05:04
[…] I insist on describing PSST as RBC without stupid microfoundations, this is also a win for PSST? Not to mention that Sumner is confused by the events. This is an honest question: how is this a RBC, anyway? What was the shock? Traditional supply-side […]
5. February 2012 at 06:13
Vuk, I don’t think velocity is a very reliable indicator of labor market changes, especially at zero rates.
dwb, I agree that during the early stages of recovery there is often a disconnect. But I’ve never seen anything even close to this.
Justin, I agree with your point about the long run average, but I still say I’ve never seen such a weird 12 month period in my entire life. This still should be showing typical “recovery” type correlations, and it’s not.
B, I agree that we shouldn’t put much weight on this number for policy purposes–indeed I think the Fed should ignore the unemployment rate and focus on NGDP.
5. February 2012 at 09:34
But wasn’t the 2009Q2-2010Q1 period also a particularly weird 12 month period? The massive decline in labor hours despite trend growth. The last five quarters (+2.6% business output growth, +1.7% hours growth) doesn’t seem all that ridiculous to me, but the huge disconnect in 2009 does.
http://www.bls.gov/lpc/
As a quick anecdote, I work in corporate finance, and our team expanded by 3 (from 8 to 11) last year. We started talking about adding new people in late 2010, a year prior to that we would have been looked at as if we had two heads if we had asked to add 3 full time employees. We had to interview >50 people to find 3 we liked, and they all started in 2011Q2. The rate of RGDP growth or even corporate profit growth wasn’t all that important other than it had been expanding for awhile and it didn’t look like it was about to crash.
7. February 2012 at 06:23
maybe – but some of it might be an artifact of the seasonal adjustments. I hope the below FRED graphs work – i’ve put 4 periods on the graph. Honestly, not seeing anything that makes me think the historical relationship has changed. It’s pretty noisy period to period, and going back to the 90s we had some strong deviations as well (the 2008 recession deviations were eually large, but to the downside instead).
http://research.stlouisfed.org/fred2/graph/?g=4Re
http://research.stlouisfed.org/fredgraph.png?g=4Re
7. February 2012 at 07:19
Justin, Why do you use business output, rather than total output?
dwb, It can’t be just seasonal adjustments, as the November 2010 to November 2011 change was what I first noticed, and blogged about 2 months ago.
13. March 2012 at 05:24
[…] Smith has interesting posts on related questions. Scott Sumner wrote an early and important post on the same topic. His bottom line was this: So what are the likely […]