The new Brookings paper on the productivity slowdown

by on March 9, 2016 at 12:43 am in Data Source, Economics, Uncategorized, Web/Tech | Permalink

This piece (pdf) is by David M. Byrne, John G. Fernald, and Marshall B.Reinsdorf.  It argues that the productivity slowdown is real, and not the result of mismeasuring the value of information technology.  Here were some of the newer bits for me:

Adjustments to equipment, software, and intangibles imply faster GDP growth but also faster input growth (since effective capital services are rising more quickly). After adjusting hardware and software, the aggregate TFP slowdown after 2004 is modestly worse. Adding additional intangibles, as in Corradoet al. (2009), works modestly in the other direction, so in our broadest adjustment for investment goods leaves the 1-1/4 percentage point TFP slowdown little changed.

And later they restate the point in more general terms:

…we highlight here the conceptual reason why it is hard for capital mismeasurement to explain the past slowdown in TFP growth: It affect inputs as well as output in largely offsetting ways.

Note that there are some unmeasured productivity gains from fracking:

…fracking allow[s] access to lower “quality” natural resources is imperfectly measured. A back-of-the-envelope calculation suggests that true aggregate labor and TFP growth might be 5 basis points faster since 2004.

Outsourcing however cuts the other way:

…the import-prices declines from offshoring are largely missed. This led to an understatement of true import growth in the late 1990s and early 2000s (the time of China’s WTO accession), and a corresponding overstatement of perhaps 10 bp in growth in output, labor productivity, and TFP.

That makes three very good papers in the last few weeks, by very reputable economists, coming from different directions, but all establishing more or less the same conclusion.  My discussions of the other two papers are here and here.

So will this myth finally die?

1 a6z March 9, 2016 at 1:32 am

Will this myth finally die? Of course not. It is valuable to certain people to keep alive.

2 So Much For Subtlety March 9, 2016 at 1:41 am

So the next question should be whether there is a conscious or unconscious political bias in the measurement. This is inherently inaccurate. People are making a lot of assumptions and estimates. It would not be hard for some people who want to see a rosy picture to see a rosy picture.

The electorate certainly seems to think things are not so rosy and their views may be the last word.

3 tokarev March 9, 2016 at 4:35 am

Keep fighting the good fight, Tyler. The IT moguls will no doubt keep pushing their meme that they are producing loads of “hidden growth.” It must be a blow to their egos to have to recognize their generation has not been able to match the rate of tech progress the previous generations (going back to the 19th century, actually) managed.

The problem at this point is that our institutions are addicted to growth. Growth is over.

4 anon March 9, 2016 at 8:46 am

The IT moguls didn’t know enough economics to back off and say “yes we have tech progress, but no that may not always translate into GDP growth.”

I think you made the opposite error, didn’t you?

I think you asserted that without GDP growth there is no tech progress. That is, “progress” is fungible and convertible to cash.

5 derek March 9, 2016 at 9:10 am

You can have tech progress without GDP and productivity growth.

Take advertising. Smart advertising is geared to optimizing the use of capital resources; if you have for example a crew of service people, by advertising you can optimize that resource by fixing a/c units in the winter. Or try to smooth out tire purchasing patterns so everyone doesn’t show up the first day it snows. Etc.

Radio, newspapers, flyer delivery, etc. were all excellent for this. You could put out a special offer, time limited, target a specific market segment, area, whatever. The costs were high but applied intelligently it made more productive use of resources.

Other than radio, this stuff is almost entirely gone. The technical advancements have shrunk the advertising world and remove productivity enhancing niches. The simple layout and design of a flyer is replaced by the necessity of having someone skilled handle online media reputation management or rankings. Something cheap became expensive and less effective as a means of allocating resources.

It hasn’t all shaken out, and a good part of the necessary infrastructure is almost there, but not quite. Localized is close. But the vaunted speed of technological advancement isn’t there; yellow pages in Canada collapsed half a decade ago, last summer was the first year that Google search actually began replacing the function on a local level, and I expect another half decade before they replace the functionality and productivity optimizing of paper.

6 Anna March 9, 2016 at 4:44 am

I’m mildly amused that we discuss “productivity slowdown” when productivity is actually growing, and only the growth is slowing down.

7 Brian Donohue March 9, 2016 at 7:37 am

+1. Pessimism = gravitas.

8 Dylan March 9, 2016 at 8:48 am

I’m amazed we discuss a “car slowdown” when it’s still moving forward, only its acceleration is negative.

9 Lord Action March 9, 2016 at 1:03 pm

You’re confusing the first and second derivative.

10 Dylan March 10, 2016 at 9:07 am

No, she is. Position of car = output, velocity/movement = productivity growth, acceleration = rate of productivity growth. In both cases “slowdown” is absolutely the correct terminology.

11 Urso March 9, 2016 at 8:57 am

Whether you want to call it a slowdown or whatever is not addressing the central point. The central point is that SV is blowing smoke up peoples’ asses about these incredible benefits that they’re creating for everybody, if only you could figure out a way to measure it. But those invisible benefits probably don’t exist.

12 anon March 9, 2016 at 9:03 am

That’s wrong. SV is split between the Second Machine Age folk who see digital tech as mainly jobs destroying, and the .. actually it is market fundamentalists who believe that markets respond to anything and create full employment again. The second group are technologists (Andreessen) but they are practicing economics when they make such claims.

13 anon March 9, 2016 at 9:08 am

Perhaps this is still an under-appreciated point: “jobs destroying” and “productivity enhancing” are not equivalent terms. Like tech/market growth, sometimes yes, but sometimes no.

14 Brian Donohue March 9, 2016 at 9:34 am

No, Urso and anon, I believe you are missing the larger point. Of course, productivity increases kill jobs. Look at how many people we employ in agriculture, or steel, compared to 100 or 50 years ago, respectively. Higher productivity is the reason the smaller number of people who remain in those fields make more money.

Productivity increases also frees labor formerly employed in less productive activities to find more productive activities, which is why it is not necessarily “job destroying” overall. This is creative destruction, it’s often very painful, and it’s another thing the perma-pessimists around here grouse about no end.

Apart from what SV blowhards are saying, productivity continues to go up, just not as fast as a guy in a university would want it to. This whole frame is amusing.

Besides, the labor market hasn’t been particularly tight for most of the last decade. Wouldn’t we expect productivity increases to be slower under such conditions anyway?

15 anon March 9, 2016 at 9:40 am

I think you missed my meaning in “not equivalent terms.”

Consider an automated donut machine. If I invent one and sell it to every doughnut store, jobs go down, productivity goes up and (depending on the ratio of my income to the former doughnut makers) GDP adjusts.

But that’s only a first order affect. Everything depends on what happens next to the doughnut market. If it stays static, my jobs-destroyer probably reduces GDP (why else would it be bought but for lower costs?). If on the other hand, cheaper doughnuts mean that everyone must have them every day, stores must be built, smiling wait staff must be added.

Or has been mentioned, Craigslist and the Newspapers. Hard to find a “productivity” story there that enhances growth.

16 anon March 9, 2016 at 9:47 am

And the data is in, right? Workers displaced in the last 20 years have not found equivalent wages. They’ve downgraded.

Are you with Andreessen, preferring quasi-religious belief to data?

17 derek March 9, 2016 at 9:22 am

But it isn’t. It isn’t growing.

That is the point, and it is obvious to me. I don’t know how many times I’ve had to deal with a ‘productivity advancement’ in either a product or a business that ends up consuming hours of my time. Some fancy system that essentially transfers the labor costs from my customer to me for accounts payable. Someone should apply a class beating in business college every year for cockamamie schemes like that.

There are one time amazing advancements that have happened. CAD, fax machines, etc. I love my phone since I can take photos of equipment tags, installation details and the like, share it with the crew, look at it wherever I am. Wow. I’ve been doing that for half a decade now. The gains are already priced in, have been for a while. Now what?

18 anon March 9, 2016 at 9:26 am

We are still on the upward ramp of “with YouTube I can do anything.” I read last week that some high percentage of plastic surgeons had used procedures they learned on YouTube. Me, I just learn how to fix the shower (GDP destroying do-it-yourself!)

19 rayward March 9, 2016 at 6:12 am

Of course, it’s possible that Silicon Valley will produce large-scale productivity growth sometime in the future, when Jeff Bezos digitizes every home ( and offers routine space flights for everyone ( And it’s possible that pigs will fly. More likely will be an economy without jobs ( The strength of the American economy was its strength, the male variety. Shifting productive capital to China, India, and other places may have resulted in lower costs, lower taxes, higher profits, much higher pay for American executives, and less direct pollution of America’s air and water, but for how long. Based on experience, not very long. Even shifting the pollution to China, India, and elsewhere was only a short-term benefit for America, as the polluted air and water there has adverse consequences here, and everywhere. The hand-wringing by American executives and politicians over pollution of air and water in China is hypocrisy of the highest order: for whom do they think China is manufacturing those goods, Martians? Maybe Trump should adopt an environmentally-themed slogan for his campaign: Build the Wall and Save the Environment (and the World).

20 B Cole March 9, 2016 at 8:28 am

Well…do these studies take into account pollution controls?

Shale oil is interesting, and I see that it is mentioned. But one could see a lot of resources being devoted to produce oil, decreasing productivity (as measured) compared to earlier eras.

21 prior_tests March 9, 2016 at 9:30 am

Come now, the idea of energy returned on energy invested would just mean that the productivity of East Texas oil fields has been sliding for years, making that whole Saudi America dream pure fantasy. Using actual measurements with actual empirical values should play no role in such debates, of course.

22 JWatts March 9, 2016 at 12:44 pm

Your post makes no sense. Oil Fields have a high energy returned on energy invested. Otherwise they won’t be profitable to extract from.

Once again, you make a nonsensical American bashing comment.

23 tokarev March 9, 2016 at 4:55 pm

>Oil Fields have a high energy returned on energy invested. Otherwise they won’t be profitable to extract from.

Technically it could still be profitable. There’s a premium for energy that can be used conveniently in internal-combustion engines. A gigajoule of energy stored in gasoline is worth a lot more than a gigajoule stored as woody biomass.

24 B Cole March 9, 2016 at 8:40 am

Add on: According to FRED, output per hour in US manufacturing has doubled from that long ago and forgotten time known as… 1995.

Not so bad. There is a leveling in recent years, but I suspect that has more to do with very slow economic growth than with a loss of innovation. Output per hour slows during recessions…we have the central bank noose on now, long term.

25 derek March 9, 2016 at 9:32 am

Two points.

A local automotive electronic component manufacturer has increased their plant productivity substantially. They have a few people watching machines crank out the pieces. They box them up, ship them to mexico where the labor intensive aspects of the assembly are done. Discrete aspects see substantial productivity improvements, but the net productivity improvement is far less.

A manufacturer that I deal with has optimized their production, inventory, business processes measurably. A large manufacturer whose numbers show up on national surveys, and when they do something it moves needles. I’ve had two occasions where my customers whose processes depend on the products have had serious productivity declines including shutdowns when their unbelievably efficient inventory management meant we had to wait for a production run of the product to get one. The choice was wait, or spend 5x more to not need it.

Damn lies and statistics.

26 Ray Lopez March 9, 2016 at 9:13 am

But TFP is itself inherently flawed, see the last thread and the “Cambridge Capital Controversy”. Put simply and crudely, TFP is a residual that is biased against intangibles that have no historical basis in the data series in question.

27 Nathan W March 9, 2016 at 10:35 am

TFP is indeed a residual. For some reason, throughout my studies of economics no one really seemed particularly interested in that. A commodity price collapse will measure as a TFP improvement. Union busting and driving down wages will reflect as a TFP improvement. But neither of these is what economists purport to be speaking of when they discuss TFP.

28 William Cotton March 9, 2016 at 11:47 am

Here in San Francisco those intangibles don’t help when it comes to paying $3,500 a month for a 1 bedroom apartment. Perhaps this economics hooey is just too darned pragmatic!

If you work for Facebook, you can afford to pay $3,500 a month for a 1 bedroom apartment. If you sit around all day using Facebook, you’re not going to make enough to afford $3,500 a month.

If you sit around reading books all days, you also can’t afford $3,500 a month, unless you also happen to write and sell a lot of books as well.

However, Facebook is not a marketplace for intellectual property. If you publish a post to Facebook that generates a lot of “likes” and “shares”, you see no direct profits, nor is your post involved in any sort of market transaction. There’s no price discovery mechanism for your Facebook post because while these “likes” and “shares” will drive ad impressions and interactions, these profits are never associated with your Facebook post.

So unlike sitting around and reading books all day where you at least have the option of writing and selling a book, sitting around and scrolling on Facebook all day leaves you with no way to see any profits from publishing a post to Facebook. If you want to make money from Facebook you need to be doing things like writing and selling books. That leaves Facebook as nothing more than a marketing platform that sits blindly on top of the unrealized potential of intellectual property assets that could otherwise be traded in a public marketplace.

tl;dr Facebook is unproductive because it is a marketing platform and not a marketplace.

29 Nathan W March 10, 2016 at 6:06 am

A friend of mine recently spent a few hundred dollars in targeted Facebook impressions, where you create unique content and it appears in the timeline of the targeted audience, and got a couple thousand dollars in business out of it. The key is good content related to the specific interests of the target group, and returns on marketing expenses are absolutely quantifiable.

30 William Cotton March 10, 2016 at 8:14 am

I’m not saying that Facebook can’t be a successful marketing platform, rather that it doesn’t promote as much economic activity as a marketplace for intellectual property.

The content that your friend made is basically worthless outside of the purpose of marketing. It’s not valuable content in the same way that a book is. How do I know this? People will actually pay for a good book but will very rarely pay to read marketing material. Marketing material is something that other Facebook users are forced to be subjected to in exchange for a free service.

When I say unproductive I don’t mean that you can’t measure the success of a marketing campaign, I mean that overall, whatever people are doing on Facebook is unproductive. Most people who use Facebook are not running marketing campaigns. They don’t see a return on any of the intellectual property that they publish on Facebook.

The actual economic activity that is facilitated by Facebook happens outside of Facebook. Your friend only used Facebook to drive the sales of some other good or service. Marketers could just as easily use some other means with which to target an audience. They could explore trade magazines, trade shows and making stronger personal connections with influential writers and other people with a big audience who might be interested in a product. All of this is healthy and productive and captured by economic indicators like GDP.

Marketing material is something that you pay to publish. A good book is something that you get paid to publish. Marketing material can only promote existing products. Intellectual property treated as a capital asset can promote the creation of new products. If you write a song it could be recorded by 50 different musicians and sold on 50 different albums. With some success you could use the promise of future profits of that song as collateral for securing a home loan. The same cannot be said for marketing material.

31 Brent March 9, 2016 at 9:40 am

Most of the impressive technology advances in the last decade have not been focused on increasing growth, but focused on increasing the perceived quality of leisure through smartphones. The companies that achieved massive wealth from this trend have leveraged their political power in a way that damaged the incentives for Americans to innovate in the direction of hard sciences and cutting edge research that will lead to real technology advances and productivity growth. Until there is a reward system back into place that incentivizes hard science over app development, I’m not sure that we will see returns to prior productivity growth again.

32 anon March 9, 2016 at 10:04 am

A man standing athwart a gold rush and yelling “stop!”

33 Tom Warner March 9, 2016 at 11:25 am

I don’t really believe in TFP as a successful measure of what it’s meant to measure, but straight labor productivity has been low since 2011, so let’s talk about that.

There are two arguments about technology advances being undermeasured that are absolutely valid.

One is that technology is becoming so much cheaper so quickly that a Fisher index of real output underweights it. In other words, if you look at how much better 2016 tech is relative to 2015 tech weighted by 2015 relative prices, the gain isn’t so great, but if you look at how how much better 2016 tech is than 2015 tech weighted by 1990 relative prices, the gain is enormous.

Another is that productivity doesn’t measure volunteer labor which in the information sphere has been made much more plentiful and easy to access. Since prices are the only way we know to objectively measure value, there is no way to confirm or refute this kind of argument.

34 Tom Warner March 9, 2016 at 11:29 am

As for non-tech factors of the recent low labor productivity, the rapid pace of retirement of high-productivity boomers is surely a big one.

35 rayward March 9, 2016 at 12:07 pm

Not since the circus came to town have people been so impressed as they have been with all that technology produced in Silicon Valley. I don’t know what’s more absurd: people being so impressed by the circus and Silicon Valley or economists debating whether the value of the time spent by Americans on their lard ass in front of a computer screen should be added to GNP.

36 tokarev March 9, 2016 at 5:00 pm

>economists debating whether the value of the time spent by Americans on their lard ass in front of a computer screen should be added to GNP

If only we could figure out how to tax people wasting time on the internet, our budget deficit would be history.

37 Scott M. Koser,CFA March 10, 2016 at 10:53 am

My concern on all of the discussion of the slowdown in productivity growth is this, don’t we have a large bank of unrewarded (in terms of salaries or wages) productivity growth from the past thirty years? Profit margins and share of GDP to capital (versus labor) are at all time highs. The only way to normalize salaries and wages from here, mathematically, looks to be negative productivity growth. IE, pay more wages and salaries for the same output, which sounds insane.

That appears to be Janet Yellen’s real goal, just look at the numbers from 4q2015. Strong job growth, weak GDP growth, negative productivity growth.

Interesting times. Too interesting.

Comments on this entry are closed.

Previous post:

Next post: