Productivity and Unit Labor Costs

Revised productivity figures were released today. The revisions were driven by a large upward revision in labor hours. This had two effects. First, with hours revised upward, output per hour - our measure of productivity - declines. Second, with a decline in productivity, unit labor costs increase. Thus, the other side of this is that with higher labor costs, output prices are more likely to be reset upward. This brings about inflation pressure making the Fed more likely to raise rates at renewed signs of inflation, and less likely to cut rates if signs of weakness show up in the data. (To be clear, the revisions to productivity and labor costs won't be news to the Fed - they were widely advertised and anticipated - so this news is already in the Fed's information set and policy response).

Here are graphs of the year over year percentage change in  output per hour for the business sector (the non-farm business sector looks very similar) from 1980 and, for more perspective, from the beginning of the sample in 1947:

The general decline in productivity changes in recent years is notable (and an impediment for those promoting tax cuts as a means of stimulating productivity growth). Will the series level off at, or just under 2%?

For more see: It's Official: Productivity Growth Has Slowed (from Dean Baker who has been publicizing this since the first hints of revisions were given), The Domestic Macroeconomic Outlook (Brad DeLong), US labour costs add to inflation threat (Financial Times, free), and Productivity Revised Lower, Costs Jump (Bloomberg), among others.

Update: In response to comments, let me clarify the statement "The revisions were driven by a large upward revision in labor hours" since it's not as clear as it could be. Here's the statement from the report:

In both the business and nonfarm business sectors, productivity growth for the fourth quarter of 2006 was revised down by 1.4 percentage points from the estimates published February 7, due solely to downward revisions to output. Productivity growth during calendar year 2006 also was less than previously reported in both sectors, but the revisions to annual data were almost entirely explained by upward revisions to hours worked.

Comments

[DILBERT DOGBERT]

An eyeball glance at the second chart says to me that there was a long decline up till the 80s and I bet that was due to the decline in manufacturing as a component of GDP. The shift to a services based economy had issues with productivity. Then there seems to be some stabilization and maybe a uptrend from there on that may be due to computer tech. Just a wild ass guess from a non-economist.
What is most interesting to me is the reduction of frequency of recessions and shortening of there spans.

[dissent]

It seems the new plants are overwhelmingly going up in low wage countries. How does that effect the productivity of workers here? Isn't the US going to eventually become derelict if new investment and upgrades go largely into low wage environments?

Much of the American economy is services, and my understanding is that it's harder to increase productivity in services. So, where are we headed?

[kharris]

"The revisions were driven by a large upward revision in labor hours."

The table I'm looking at shows a 0.9% rise in hours in the revised data, vs a 1.2% rise in the initial release. It shows output up 2.5% rather than 4.2%. If we look at revisions on a y/y basis, rather than q/q, output was revised down 0.5% to 3.4%, while hours were revised up just 0.2% to 2.0%. Output was the big factor in the revisions, not hours.

Dissent,

The pattern has mostly been for jobs to move offshore from low-productivity US industries. (I'm not sure that's the case in the chip sector, though.) When low-productivity industries leave, US productivity goes up, even without the productivity of any remaining job improving. In fact, productivity in those remaining jobs has also improved, so two factors are boosting productivity. What semi-skilled workers need is capital poored into their industries, so that only one factor drives up productivity.

By the way, when you hear that productivity is the greater cause of US factory sector job loss, it is worth keeping in mind that there are two factors at work. Measured productivity is going up partly because low productivity jobs are leaving. In that sense, rising productivity is costing low end factory jobs precisely because they are going off shore.

[Mark Thoma]

kharris: I was looking at this part of the report from the BLS:In both the business and nonfarm business sectors, productivity growth for the fourth quarter of 2006 was revised down by 1.4 percentage points from the estimates published February 7, due solely to downward revisions to output. Productivity growth during calendar year 2006 also was less than previously reported in both sectors, but the revisions to annual data were almost entirely explained by upward revisions to hours worked.I should have read closer. Thanks. I'll add an update.

[Movie Guy]

kharris - "What semi-skilled workers need is capital poored into their industries, so that only one factor drives up productivity."

k, why not explain point in more detail.

[johnchx]

dissent wrote: It seems the new plants are overwhelmingly going up in low wage countries. How does that effect the productivity of workers here?

That's a good question. One of the frequently asked questions about global trade is whether industrialized countries will suffer from a "capital hemmorhage" as investors seek higher returns in non-industrial countries.

Interestingly, the facts to date don't seem to support this story.

First, in general, global financial capital is not flowing from the developed world to the developing world. Instead, it is flowing from the developing world and the non-U.S. developed world into the U.S. The "giant sucking sound" you hear is the U.S. economy vacuuming up capital from all over the the world.

Second, the Bureau of Labor Statistics' multi-factor productivity statistics show that private business in the U.S. is consistently becoming more capital intensive, rather than less. For instance, from 1995 to 2005, capital services per hour of labor (for all private business) grew by 41%. Over the 1987-2005 period, capital per labor hour grew by 62%.

The multi-factor productivity data also show that changes capital intensity actually contribute a smaller share of productivity gains than one might guess. In 2005, for example, total output per hour grew by 2.6%. Of that, 0.7% could be attributed to increased capital intensity; another 0.1% was attributable to changes in workforce composition (i.e. gains in education and experience). The remaining 1.8%, almost 70% of the productivity gain, is more or less unexplained -- the result of organizational improvements, technology, off-shoring less productive tasks, and who know what else (including, of course, possible measurement error).

So the good news is that the visible story -- foreign owned plants being built in industrializing countries -- isn't the whole story. And if opportunities to invest in developing countries are exerting downward pressure on capital investment in the U.S. (and it seems like they must, in some degree), other factors would seem to be offsetting the effects.

At the same time, there is room for concern: it seems to me that we're allocating an excessive share of our (and the world's) savings into housing stock and into consumption of government services (e.g. the Iraq war and other deficit-financed enterprises). It is those uses that are pulling capital away from domestic business investment, IMO.

[ken melvin]

A chicken in every pot and 2% forever.

[calmo]

Not enough chickens and 2% could be so generous for the next few quarters.

[calmo]

No 2 ways about it, when I look at those graphs, I know that measured productivity is about as convincing as reading a measuring tape during an earthquake. At the very least I want to see a much thicker line. [Hey, they revise, I get to use a brick-like pencil.] Yes, an increasingly thicker line from 1980 when financial productivity starts to muddy the waters.
And another foul-mood thing: just how come bonuses paid to some employees are counted (serious bonuses to managers not tallied) and not some new rule implemented about ex-free-pizza wages?
Very last foul-mood thing: Imagine 2 identical houses that got built with identical hours, 1 sold last year for $400,000 and the other this year for $300,000. Do the productivity stats adjust the product (hedonically?) so that the intuitive notion --that the activity represents the same productivity, is preserved?

[Martin]

JohnCHX,

Stephen Roach of Morgan Stanley has charted the capital inflow you describe - quoting from the top of my head, capital's share of global GDP has increased from 10 to 16% over the period from 2000, while labour's share has shrunk from 56% to 53% over the same period.

There has never been a better time in history to be rich.

The real question over these figures, or indeed any such figures, is how they might affect general economic confidence. Last week proved that when Shanghai now sneezes, the rest of the world might not catch a cold; but it will develop a shiver.

And when the cold does come, macho posturing of the kind indulged in by Gordon Brown, the UK's curent Chancellor of the Exchequer and likely next Prime Minister, concerning whose productivity's biggest isn't really going to help us.

[Bruce Webb]

How do you possibly make policy decisions on a data series that can be thrown so over the map as productivity has over the last four quarters? What if anything is actually being measured here and how?

My faith was totally shaken in Q4 2005. Productivity had been clicking in at a nice steady 3.0+ rate for years, predicting that it would come in north of 2.0% on an annual basis was as easy as falling off a bicycle. Suddenly it screached to a halt. Productivity in Q4 2005 came in at zero giving a total 2005 figure of 2.0%, oddly enough right in line with Social Security Interemediate Cost. Then Q1 2006 came in at 5.6%. Now I look at a data series that goes 3% 3% 3% 0% 5.6% 2.8% and think there is more likely than not some problem in the reporting, some part of growth that may have actually occured in Q4 2005 (reported at zero) somehow got reflected in Q1 2006 (5.6%? somebody recovered from that New Years hangover pretty quick). Then hidden in the Q2 BLS release of productivity was the news that they had changed the measure altogether, altering the way they handled light truck production, which had the result of slicing .3 points off of 2002-2004 productivity.

This is insanity. You have a data series that over the course of a month can be adjusted down by 40% (because I can't take "revised down by 1.4 percentage points from the estimates published February 7, due solely to downward revisions to output") any other way and you have to conclude that drawing any particular inference about how the economy is actually performing from this is crazy. The Romans probably got better results examining the entrails of chickens by the Augurs.

Now I have my own particular paranoias, well known to the cognoscenti on Economist's View, but this is the second Q4 in a row when spectacular revisions to reported productivity came in at particularly convenient times.

You can have a measure that can be adjusted down by 1.4 points in the four weeks since Feb 7, 2007 or you can have policy prescriptions that depend on identifying the level of that measure in 2013 and projecting it over the Infinite Future Horizon and proposing policy changes based on that. I don't see how you can have both.

Thank God I have never claimed to be an economist because it seems that you all are being thrown a totally crapped data series and being told to make sense of it. This isn't Economics, it barely rises to the level of Santeria. The economy did what it did in Q4, data elements like income tax deductions and FICA contributions, I mean actual real hard dollar figures, actually rolled in. Surely someone could backward compile from this some real numbers that didn't vary by 40%.

Were there actual revisions to the data? Was the reporting in some way erroneous? Who exactly decided on this "revision to output"? Why, how and based on what?

I don't want to cast aspersions on public servants. I spent my entire career until last July drawing a public check. But I have participated in meetings where senior management discussed when to start the clock on permit issuance. Did the permit take six weeks or two weeks? Well if you start the clock based on external agency approval, when the Health District signed off, your numbers improved in a hurry.

Sorry, I just don't trust the productivity series. It has been whipsawed too many times in magnitudes way out of whack with its underlying number over the last six quarters for me to give it any credence at all. "Did we say 3.1%? No we meant 1.7%" over a four week period is not an economic measurement of anything. Something is rotten in the state of Denmark. It is nice to believe that every revision to a data series is biased in the direction of accuracy, that is you are better measuring whatever it is that you are trying to measure, but this is ridiculous.

Perhaps people who actually know something can set me straight on this one. I am just a guy reading a chart. It just seems that anyone who deploys a number that proves to be correct with an error margin of plus or minus 40% is not reporting anything at all.

Bueller? Bueller? Bueller? Somebody help me out here.

[Movie Guy]

Productivity? Why not focus some serious attention on the operations being offshored?

There could be an ongoing effort to identify by corporation the offshored operations and numbers of employees fired/displaced on a monthly basis. Further, the record of inbound imports by commodity group and level of technology could be tracked.

There is no effort to measure the production valued-added losses nor the dollar value of such losses occurring in the U.S. economy. Meanwhile, we continue to focus on productivity numbers. This is like looking at half of the issue.

[Dean Baker]

Quarterly productivity data are close to useless. There is a huge amount of error and imputations that drives the numbers. Annual data are respectable, since the errors tend to be offsetting over the course of a year. The numbers that we use to get productivity are output and hours worked. If we don't have a reasonably good estimate of these variables, at least on an annual basis, then we really can't say very much about anything to do with the economy.

[Real Person from the Real World]

Offshoring of programming - most companies just farm out the less important stuff. Important stuff is still in-house. Still worried about all those programmers from India coming over here? Got news for you.... a lot of these guys may go back. India is experiencing a scarcity of midlevel & upper IT people. Some will probably be returning, taking with them, all the knowledge they got on the job working here. How will that affect productivity, Hhmmm? We need to get those of us born here better educated to take over. That means we need entry-level IT jobs. Education, Inc., needs to stop gouging on overpriced textbooks, and credit hours and education costs that puts a huge debt burden on people's backs for life. Maybe more help to our own people, less lavishing it on foreign students.

says...

P.S. I have nothing against foreign students, but I don't have a mother country to run to, should things go sour here for me.

[spencer]

The main reason 4th Q productivity was revised down was the downward revision of 4th Q real gdp.

It is not that the data is so bad. Rather, we report the data before all the information is in. Originally, the first gdp report was just an internal BEA series. But it was leaked so badly that BEA decided to publish it.
Maybe they should have decided to quit calculating it.

The dominant factor driving profits is the spread between unit labor cost and nonfarm prices. this report implies that profits are under severe pressure and implies that the main reason for the recent market downturn may be a sharp downward adjustments to investors near term eps expectations rather then some esoteric international financial development.

Some of the best people on productivity just put out a new report that should be worth reading.

2) “A Retrospective Look at the U.S. Productivity Growth Resurgence,” by Dale W. Jorgenson, Mun S. Ho, and Kevin J. Stiroh
It is now widely recognized that information technology (IT) was critical to the dramatic acceleration of U.S. labor productivity growth in the mid-1990s. This paper traces the evolution of productivity estimates to document how and when this perception emerged. Early studies concluded that IT was relatively unimportant. It was only after the massive IT investment boom of the late 1990s that this investment and underlying productivity increases in the IT-producing sectors were identified as important sources of growth. Although IT has diminished in significance since the dot-com crash of 2000, the authors project that private sector productivity growth will average around 2.5 percent per year for the next decade, a pace that is only moderately below the average for the 1995-2005 period.
Read the full report:
http://www.newyorkfed.org/research/staff_reports/sr277.html

[Bruce Webb]

Thanks Spencer.

How could the preliminary data be so off as to require a 43% downward revision between preliminary and final? I don't understand productivity particularly but I follow it pretty closely as a pure number series in relation to Social Security and absent some serious and visible jolt I have not been aware of this level of revision, how normal is a 40% change between preliminary and final?

To the non-professional the number series has looked rather odd for the last five quarters, going to essentially zero in Q4 2005, then roaring back at 5.6% in Q1 2006. Then dropping back to (I believe) 2.8% in Q2, where it was accompanied by a change to the measure itself.

BTW that 3.0% Q4 number displayed on the home page of the BLS just a couple of weeks ago http://www.bls.gov/, it looked awful official then, sure enough it reads 1.6% today.

And I still find it odd that 2005 ended up with that disappointing 2.0% number and yet the Trust Fund ended up with a substantially larger balance ($13.9 billion more) than was anticipated by Intermediate Cost.

Well the 2007 Social Security Report is due out in a couple of weeks (though unaccountably it was delayed to May last year) and Q1 productivity should come soon after. It will be interesting to see how they shake out.

[calmo]

As far as I know there has always been (ok, in the last decade) an "advance", a "preliminary", a "final" and then sometimes a "revised" or 2 on that GDP stat.
[Of course, the denominator 'hours worked' is a major stumbling block for those of us who can't dislodge 12M illegal aliens from their memory bank.]
As far as I have learned reading the SS reports, the productivity numbers are not merely lifted from the latest BLS stats but compiled in packages of periods (peak to peak) and averaged. [Quite a lot of past data goes into making that 75yr projection, yes?] Since we have not come to a peak in this cycle(?) the very latest number is not firm.
The fact that they do not detail this derivation tells (very cynical lately) me that they have no interest in educating the public, only in commanding compliance. [The entire w administration m.o.] We are supposed to trust them --no matter the SS executives not making any squeaks about the "empty cabinet full of IOUs". [Isn't this like a major dis to your life work of ensuring that this fund is properly managed?...and by political hacks who see only a few years horizon?] It is called the Social Security Trust Fund because we are supposed to take Lucy at her word that she won't take the football away again.
[Actually I want to be Lucy. I'm tired of being Charlie Browned. Isn't everybody?]

Back to the current handling of some important items in the CPI basket and hedonic pricing about which I am just a little under-educated. The latest computer is twice as fast and twice as powerful (should that be different) than your old computer. And recorded as twice as much product despite costing half as much. This accounting inflates GDP esp in the 90s when we were all threatened with being left behind...but it is a small component of GDP now.
Unlike housing which may undergo similar price revisions in the coming quarters. Not many people think you'll get a much better house today than a couple of years ago. Not likely the OER can be penciled in higher because it is twice the house. If recent inventory hangs up for lack of qualified buyers, rents may actually come down along with those house prices which the officials ignore anyway. Will the housing component of GDP growth actually have a negative contribution to GDP like Imports or will there be a revision in the way OER works to preserve the status quo?
Ok, my half-baked thought is that concealing the impact and importance of housing in GDP via proxies and perhaps other devices, might do wonders on the way up, but on the way down...a possibility that we see sub-European GDP numbers.

[Real Person from the Real World]

Still arguing about "non-farm" labor, GDP, and units of productivity. No one worried about all those programmers from India that may or rather probably will go back because India is experiencing a scarcity of midlevel & upper IT people? That they may be taking with them, all the knowledge they got on the job working here? So, n o one figures that will affect productivity even a little? You are ALL ostritches with your heads in the sand.

[Real Person from the Real World]

A repeat from another page:
We seem to live in a society where everyone is always looking for someone else (as long as it is NOT them) pays for something. People do not want to pay a fair wage, so give them a tax credit, that way they don't have to pay, but they can talk about how socially minded the organization is! Meanwhile, the disadvantaged worker is working, but still has difficulty making ends meet. The thinking that needs to be changed is that a worker is a comodity. I asked my immigrant employer about whether he would ever sponsor a needy child's education in his homeland. He said that was why they hired maids. Well, that's the new global economy for you.

[ken melvin]

We must fight for America, but to hell with all Americans except me and perhaps a few of mine. This extends down through the lower-middle-class.

[calmo]

A call to arms (wings?) has been issued, fellow ostriches.
You are ALL ostritches with your heads in the sand. [nevamind that if our heads were in the sand, we could not hear this message. nevamind we might get the impression that the messenger also has his head in the sand in ignoring this little detail. nevamind he is reaching out to us with this rescue-gone-bad.]
Arise, you feathered beasts, you descendants from the mighty dinosaurs who had other survival strategies than sticking their heads in the sand...less successful ones it appears.
I can get to your post Real Person, I can...and I don't even like Ostriches but, well, I am powerless to your passionate imperatives...at least in the short term (of course, you're right: head under a couple of feet of sand. At least!)
You figure the aboriginals here make that same distinction you do about immigrants? Most of the white folk concede that this land is full of immigrants and although a few may go back to Whereveritwastheycamefrom, most remain here because living conditions and working conditions are better. Those suitcases of knowledge that they collected here, that are about to leave...are swamped by incoming (not exactly empty) suitcases...just ask Bill Gates.

[Richard Schweitzer]

One element that seems to be missing attention is the use of "overtime," and how it gets tabulated in the BLS systems.

The degree of use of overtime (more pay per hour) is of course reflected in the total wage cost equation, which has been rising.

Is it possible that BLS inputs overtime as , say, 1.5 hours, when it is actually the same number of hours paid at 1.5?

From exposure to studies made for other purposes, overtime wages seem to be on an increasing gradient, generally indicative of a shortage of adequate skills, or need for momentum in production systems.

[Bruce Webb]

Speaking of odd data series can someone tell me how you get this one from that same BLS release.

"Hourly compensation grew 8.2 percent in the fourth quarter, and when the decrease in consumer prices was taken into account, real hourly compensation grew even faster, 10.5 percent. Real hourly compensation had fallen in the second and third quarters of 2006, following an 11.0-percent gain in the first quarter."

11%/neg/neg/10.5%. Did this whipsaw actually show up in the Real World? I am disliking these derived measures more and more. They may be useful technical tools in the right hands but to an outsider, a sharp decrease in productivity, combined with a spike in hours worked, itself combined with a spike in real hourly compensation would seem to mathematically combine to a huge squeeze on profits from Q3 to Q4.

Or is it all the effect of executive bonuses?

[ken melvin]

What does productivity per se (if there is such a thing, but there must be if there's a change, huh?)look like over the past 50 years?

[calmo]

Richard, I know plenty of people that work 50+ hrs and get paid for 40. It is an understanding that you might get a bonus and if you don't work the extra, you will be first in line for terminations.
The hours worked under-estimate the real hours logged IMO and hence, productivity overstated. A phenomena more prevalent today than 10years ago, IMO.

Then there are the illegals who make this entire detail moot.
Bruce, I saw those stats too and found them incredible. Bonuses was the culprit apparently, but lets be clear that executives and management compensation in general are not tallied by the BLS. [Check the job description of CEO and marvel at those modest hourly rates: it is because they are managing no employees.] Did the financial sector, for instance, dispense bonuses to secretaries of this order of magnitude to affect the total wage picture ~10%?
I don't know any ordinary workers who got these kind of bonuses and I don't think the financial sector has that kind of impact on the larger picture.

ken, you don't like the second graph in this post? Maybe it's the "productivity per se" and you just refuse to recognize 'Output per hour of All Persons' as it. Well the BLS has more specialized varieties if you are fussy.

[Real Person from the Real World]

Hey Ostriches! If India takes back NRIs, and the Visa quota stays up, where are they all the incoming IT suitcases Billy Gates wants, coming from? China? Korea? In India, English was almost a second language. While many global citizens speak it, they may not be as good at it. We have "interesting" times ahead.

[Real Person from the Real World]

Oh,; and don't forget, India may develop into a formidable competitor. We may scoff at the Indians that come here, and their schools back home, but, like all humans, they too learn from their mistakes, and they will WILL get better at things. Importing cheap IT talent from overseas, and letting our own talent fade away will cause problems in the long term.... As I said, we have "interesting" times ahead, and I do not mean prosperous times.

[ken melvin]

Change. Always the change. I wish to see that that has changed.