August 15, 2003

More Good GDP (and Productivity) News About the Second Quarter

News about the economy in the spring released since the initial estimate of second-quarter GDP growth tells us that the second-quarter growth rate is likely to be revised upward significantly--from a 2.4% annual growth rate to perhaps a 3.0% annual growth rate. This is excellent news for production and productivity: Analysts see U.S. 2nd quarter GDP upward revision: Merchandise trade data out on Thursday showed exports rose a healthy 2.4 percent in June, while in real terms, which is what matters for gross domestic product, the trade deficit narrowed sharply to $47.23 billion from $50.04 billion in May. The [trade] deficit was much smaller than that assumed by statisticians in the advance measure of GDP released on July 31, suggesting trade subtracted less from growth than the 1.56 percentage points initially estimated. The trade news comes hot on the heels of significant upward revisions to retail sales figures for both May and June. On Wednesday the Commerce Department unexpectedly upped its June estimate of sales to a rise of 0.9 percent from 0.5 percent, while May now shows a gain of 0.5 percent when it was flat before. The sudden discovery that consumers spent a lot more than first thought implied a sizable boost to economic growth.

"Second-quarter GDP will be revised higher," said Gerald Cohen, senior economist at Merrill Lynch. "Based on yesterday's retail numbers, second quarter GDP will increase from 2.4 percent to 2.8 percent, and with today's data, we believe it will be closer to 3.00"...

We had thought that the deterioration of labor market conditions in the spring was another piece of evidence that trend productivity growth is fast--so fast that a 2.4% annual real GDP growth rate is far below the economy's sustainable pace of growth. Now it looks like the spring was another piece of evidence that trend productivity growth is even faster--so fast a 3.0% annual real GDP growth rate is far below what is needed to even hold the unemployment rate stable.

I need to find an hour today to figure out how much of now-anticipated upward revisions by the BEA to the second-quarter estimate of GDP growth will also show up in upward revisions by the BLS to its second-quarter nonfarm business productivity number of a 5.7% annual rate...

I keep looking at this figure:

It used to be that American productivity growth was pretty clearly procyclical: productivity would be high relative to trend when unemployment was low, and low relative to trend when unemployment was high: periods when the labor market was deteriorating were periods when productivity growth would be slow or negative; periods when the labor market was improving would be periods when productivity growth would be unusually rapid.

The usual explanation of this productivity-procyclicality was "labor hoarding" by firms (and, in Europe, "job hoarding" by workers). Even if a business didn't have work to employ its skilled, experience, core work force fully in a downturn, it was still worthwhile to keep skilled, experienced workers on the payroll rather than lay them off, watch them drift away to other jobs and other cities, and find it very expensive to replace them with others of sufficient skill and experience when the karmic wheel of the business cycle turned in a favorable direction. We can see this pattern clearly at work in the mid 1970s in the oil-shock recession, and in the early 1980s during the Volcker disinflation: when unemployment is rising fastest the red eight-quarter centered productivity growth trend is negative; when unemployment is falling fastest productivity growth peaks. You can even see this pattern at work a little bit at the end of the 1980s and the beginning of the 1990s: there is little fall in productivity growth visible around the 1990-1991 recession (but productivity growth was already very low in the late-Reagan high-deficit low-investment years--I suspect in large part because of the drag exerted on the economy by the deficit), but there is a large productivity bounce-back during the "jobless recovery" period as employment starts to rise again, a bounce-back that quickly comes to an end. The eight-quarter productivity growth trend falls back to its anemic high-deficit late-Reagan levels until falling prices of computers and rising deficit-reduction and capital-inflow fueled investment spending lights the new economy productivity growth rocket in the mid-1990s.

And when that happens, things change. The figure below shows how the 8-quarter centered moving average of productivity growth covaries with the 8-quarter centered moving average of nonfarm business output growth since 1960. Notice the branch growing out of the top left of the scatter: that's our most recent data, the data for the moving average centered around 2000, 2001, and early 2002. Productivity growth this fast simply is not what we would expect with output growth this slow: to get productivity growth above a 4% annual rate used to require output growth at more than a 5.5% rate--not a rate of less than 2%.

In fact, the normal positive covariance between two-year moving averages of output growth and productivity growth has been breaking down since the start of the new economy productivity boom in 1995:

The picture you get, looking at the moving average data since 1995, is that the two-year moving average of output growth bounces around (depending on the state of the business cycle), while the two-year moving average of productivity growth rises in a pattern only tenuously connected to real demand growth: from less than 1% per year in 1995 to 2.5% per year in 1998 to 3.5% per year in 2001.

So what's going to happen to nonfarm business productivity growth next? We are outside the bounds of historical experience for the modern American economy, which makes any kind of forecasting extremely, extremely hazardous: something has changed to give us this productivity-output pattern, and rules of thumb and models based on past experience by definition do not take whatever has changed into account. Nevertheless...

CEA Chair Greg Mankiw said last week that the 3.5% per year real GDP growth rate that the administration is projecting should make the unemployment rate next year lower than it is today. When I look at these productivity-and-output graphs, I can't see it. Output has to grow nearly 1% per year faster than productivity in order to hold the unemployment rate steady, and that can be the case with a 3.5% output growth rate only if trend productivity growth is now less than 2.5% per year. After staring at these figures, it seems to me likely that unemployment is likely to rise with anything less than 4% real GDP growth rate over the next several years...

Posted by DeLong at August 15, 2003 09:54 AM | TrackBack


A query for Brad and others who follow productivity: To what extent are these high growth rates the products of shifts in the composition of the work force? I ask as an observer of the southern economy, particularly its "traditional" manufacturing, which lately has been going through the wringer. Several weeks ago, Pillowtex, Inc. [makers of Fieldcrest and Cannon home furnishings, among others] declared bankruptcy and shut down. It was the largest mass layoff in North Carolina history, and among other things has devastated the small city of Kannapolis, once a Cannon company-owned town. Textiles have long provided lots of (well, barely adequate) jobs, but have been (I think rightly) faulted for discouraging human capital investment in textile-dependent areas (Such places typically report numbers of adults without high-school educations that are high even by southern standards). The few prospering firms in the industry seem to be very much on the high end of the innovation scale, offering sophisticated products, flexible production, and state-of-the-art customer service. Which leads me back to my question: How much of our productivity growth comes from innovation and its dissemination, and how much from industries at the low end of the productivity-growth scale succumbing to global pressures?

Posted by: David L. Carlton on August 15, 2003 12:42 PM

Ah. The "improve average health by killing the sick" effect...

Such effects are usually small unless the unemployment rate is rising rapidly. But I haven't seen any recent estimates...

Posted by: Brad DeLong on August 15, 2003 12:49 PM

Then darn, why not fiscal policy that favors development of infrastructure to bring higher GDP growth and employment gains to take advange of our productivity growth? Why not? Why not?

Posted by: anne F&B on August 15, 2003 01:15 PM

Why not another "New Deal?" Suppose we had that rather than tax cuts. What a foolish thought.

Posted by: Anne F&B on August 15, 2003 01:17 PM

Care to substitute a manner of TVA [imagine a project] for a cut in dividend taxes?

Posted by: anne on August 15, 2003 01:19 PM

For most of the late 90's, according to this graph, GDP grew at a 4% pace, productivity at a 2% pace (roughly.) That means labor hours must have been growing at about a 2% pace. Since it is unlikely that the labor force was growing at that pace, that means increased hours for existing workers. Now, GDP is growing at a 2% pace, productivity is growing at a 4% pace. I happen to know that the employed labor force has been stagnant the last few months. Therefore, hours worked by existing workers must be dropping at a 2% pace.

Thesis: Today's differential between productivity and GDP represents a correction of the excesses in working hours of the dot-com era.

But why is productivity growth so high? Is there some new technology being incorporated? It doesn't seem so. So what, then? Well, maybe people are working fewer hours, but are getting the same amount or more done. Who has been laid off? Many tech jobs have gone away. Maybe those workers were kept around for when they were needed, but weren't needed all of the time. So the remaining ones are able to complete most of the tasks of the larger prior set, in the same or fewer hours. Fewer tech people and a lower tech budget means fewer upgrades. Fewer upgrades means less retraining of existing personnel. Less retraining means existing personnel can use their competency with the existing system to produce more - an expert with an "old" system often produces more than a novice in the "new" one.

So, slowing the upgrade cycle allows productivity growth from existing non-tech-support workers who are competent with the existing system.

OK, probably not enough to explain all of this growth. But some.

Posted by: rvman on August 15, 2003 01:23 PM

What about improvements in the retail sector (Wall Mart [IKEA])?

And what about Virginia Postrel's "nail salons"?
Creating (in the same amount of time) something the customer really likes instead of something the customer only somewhat likes - that has to count as real productivity gain (unless its fiddled away somehow in the inflation measurement).

Posted by: Mats on August 15, 2003 02:10 PM

For instance, let's say all consumers have constant taste. Specifically, I like a certain nuance of blue. Last year, I couldn't find products of this exact color in the shops.

I bought a cell phone then, with the wrong blue color, and agreed to pay USD50.

This year, producers (and distributors) improved (without using more resources) in meeting customers demand. I found and bought a cell phone with the perfect right blue color (otherwise exactly the same as last year's), and was happy to pay 100 USD.

=> Productivity growth that could easily be mistaken for inflation.

Posted by: Mats on August 15, 2003 02:25 PM

"Is there some new technology being incorporated?"

RVMAN, consider this story -- Several years ago I was lucky enough to find myself living in one of the first neighborhoods served by Cablevision's broadband rollout (not the test area, but right after). Since at the time non-techie types didn't understand what this implied for them, I was able to get an appointment right away, and they sent over a very bright and well trained young guy who spent a couple of hours fiddling around at the pole, running a cable, and testing the signal strength.

For the next three years I would experience every few months an intermittent connection problem which every time would result in the same bizarre sequence -- place call to service rep--perform modem check while on phone--told that it is working fine--repeat call to service rep next day--repeat modem check--demand house call--at house call they test signal strength--told working fine--about a week later they return to my house with three trucks and they would drive up and down the street all day checking out the other 6 customers--a week after that the repair trucks would spend a day (or days) checking out the wires and then it would start working correctly for the next few months.

And this was not only me. It happened in just this way (years later) to a couple of friends I helped set up. In one of the instances it took two phone calls and two house calls when the guy just looks at the modem and goes--oh that's the one that's incompatible (yes, they supplied the incompatible modem), replaces it and it works fine.

I don't know what happened to the guys who practically had a full time job fixing my internet service. I no longer have problems, and if there is a temporary problem I call up and the outage is mentioned in a recording. Nobody I know has any problems. They don't send anybody over to meticulously install the service, they just mail out a modem and a splitter. When somebody new joins up it costs them practically nothing because all the bugs have been fixed, and the capacity to handle the new customer is already in place.

I believe this type of pattern is quite common throughout the economy. Along with the new more productive way of doing things came a lot of workers to install and fiddle with the devices that now work out of the box and work out all the bugs in the processes. But by now we've had some time to work on streamlining the technology that before was so urgently needed that money could be thrown at it (PC support in large corporations, for instance). Not to mention that some applications of the technology just didn't work out profitably as hoped and there are economies of scale as with Cablevision broadband. Prof. DeLong and some others have suggested that with previous innovations it took decades to realize the productivity as the economy organized around the innovation but I'm not sure that this is the case here. Most companies have already organized themselves completely around info tech. In fact, they did it so fast that they had no time to worry about doing it cheaply. Actually, I remember reading an article by Krugman written at the peak which pretty much predicted that there would be fewer IT workers needed once we got good at it.

Posted by: snsterling on August 15, 2003 11:26 PM

Casting doubt on the productivity miracle:

U.S. productivity may be more mirage than miracle

Thu August 14, 2003 12:02 PM ET

By Pedro Nicolaci da Costa

NEW YORK, Aug 14 (Reuters) - For many in the Wall Street rat-race, the much-touted U.S. productivity "miracle" looks more like a mirage, and a faint one at that.

Productivity, a measure of worker efficiency, has long been the poster-child of Federal Reserve Chief Alan Greenspan. During the late 1990s boom, the central banker often hailed the improvements as the foundation for a "new economy" that could sustain soaring growth rates.

But analysts have real problems with the way this new paradigm is measured, suspecting it is both overstated and overblown.

"America's productivity saga has become the most abused concept in today's macro debate," argues Stephen Roach, chief economist at Morgan Stanley.

Productivity is a measure of output per worker and is simple enough to gauge in an area like manufacturing. If a company makes cars, the more cars its workers produce in a given period of time, the more productive they are.

The trouble arises, economists note, when the same calculus is transferred to the services sector, which employs around 80 percent of the country's workforce but whose output is not nearly as easy to quantify.

Just last week, the Bureau of Labor Statistics reported a staggering 5.7 percent jump in productivity during the second quarter. But exactly how much output the service sector produced in that period is any number-cruncher's guess.

For instance, how does one measure the productivity of a secretary? Is it the number of letters she can type per day or how many meetings she can arrange? What about a flight attendant or a waiter? There are no easy answers, nor does the government have a single standard that works across industries.

"Do we really have a clue how to measure white-collar productivity in the ever-amorphous services sector? The short answer is no -- not even close," explained Roach in a research note.


One of the most obvious flaws in the government's measure of productivity is the estimate for total hours worked. Gross underestimates are particularly striking in the financial services industry.

Economists and traders in New York chuckle wryly at the government's figures, which depict the investor community working an average of just over 35 hours per week.

"In Germany, they might. Here, it's more like 50 plus," said Jeoff Hall, chief North American economist at Thomson IFR.

Which is not to say that there have been no productivity gains at all. Technological advancements, computers in particular, have brought very real benefits to U.S. producers and consumers alike.

But even here, such strides have been most tangible in the tech-savvy computer industry itself. The average techno-peasant, meanwhile, is left to struggle with unwieldy programs, lengthy helpdesk calls and a mountain of spam.


Now that the economy is overcoming recession and trudging through a choppy recovery, the downside of productivity has become painfully obvious to America's 9 million unemployed: Highly productive firms require fewer workers to get the job done.

"You've got a lot of layoffs in the U.S. economy and two jobs combined into one," said Ethan Harris, senior economist at Lehman Brothers. "From the company's point of view that's a big increase in productivity. From the workers point of view, it's a big increase in hours worked."

Herein lies another major shortcoming in the data. They fail to distinguish between good productivity, which stimulates a real economic expansion, and bad productivity, achieved through job cuts and the hiring of more part-time workers.

Progress has other pitfalls too, such as the extra work that employees are often expected to get done away from the office because of things like e-mail, cellular phones and mobile Internet access.

"Here I am commuting back and forth from Manhattan with my blackberry, and papers that my staff is working on, and my newspaper and everything else and I'm busy churning away," said Ethan Harris, senior economist at Lehman Brothers.

"But my guess is that, since I never put in a time card at my firm, probably my boss reports me as some nominal number of hours -- 40 or 35."

Posted by: Pooh on August 16, 2003 07:22 AM

Two things:

1) I'm working (part time) as a political organizer, and it is damn near impossible to estimate my productivity growth. Yes, I still have to make phone calls and get people to show up to events, but the availability of email and websites as methods of communication to my volunteers has meant that I spend massively less time per volunteer getting them on board and keeping them informed. Mailings which used to cost me $100 and hours upon hours now cost me $0.30 in broadband access fees (amortized :)) and twenty minutes. There definitely is some real productivity growth out there, though of course it affects some sectors more than others.

2) Is the solution to massive productivity growth combined with unemployment a simple reduction in the workweek?

Posted by: Kimmitt on August 16, 2003 03:43 PM

The President's stated fact that the blame for job losses is higher productivity is the biggest joke that I have heard in a long time. The above discussion notwithstanding.

Posted by: Nick on September 5, 2003 06:32 PM

Is the productivity of labor defined as output relative to hours worked?

Posted by: Laura in DC on October 5, 2003 01:57 PM

Is the productivity of labor defined as output relative to hours worked?

Posted by: Laura in DC on October 5, 2003 02:05 PM
Post a comment