Where did the productivity slowdown come from?

by on June 24, 2014 at 2:42 am in Data Source, Economics, Uncategorized | Permalink

John Fernald has a new NBER paper on this question.  Here is the abstract:

U.S. labor and total-factor productivity growth slowed prior to the Great Recession. The timing rules out explanations that focus on disruptions during or since the recession, and industry and state data rule out “bubble economy” stories related to housing or finance. The slowdown is located in industries that produce information technology (IT) or that use IT intensively, consistent with a return to normal productivity growth after nearly a decade of exceptional IT-fueled gains. A calibrated growth model suggests trend productivity growth has returned close to its 1973-1995 pace. Slower underlying productivity growth implies less economic slack than recently estimated by the Congressional Budget Office. As of 2013, about ¾ of the shortfall of actual output from (overly optimistic) pre-recession trends reflects a reduction in the level of potential.

The ungated version is here (pdf).

1 Ian Tindale June 24, 2014 at 3:11 am

The answer’s obvious: facebook

2 prior_approval June 24, 2014 at 3:46 am

Or flickr.

3 Tom June 24, 2014 at 4:18 am

I think his timing of the productivity growth slowdown is a bit too simplistic. Productivity growth is cyclical, so one would expect a slowdown before a recession, even if there weren’t any slowing of the long-term trend.

Also, if you’re looking for the impact of the housing bubble on productivity you have to look over a longer time frame and to get the full picture you need to use net aggregates. While the bubble was on, the productivity of residential housing construction appeared high because housing prices were elevated. Then when prices fell but costs were sticky, sector productivity collapsed while elevated depreciation erased the seeming wealth created by the seeming productivity of the bubble years.

4 Steve Sailer June 24, 2014 at 5:20 am

Good thing we let in all those illegal immigrants back then.

5 C June 24, 2014 at 7:13 am

Great and relevant point.

6 josh June 24, 2014 at 7:30 am

Agreed. +1

7 Phill June 24, 2014 at 9:40 am

I KNEW IT, I saw your name and before I scrolled I wondered how you were going to bring this up.

8 Ricardo June 24, 2014 at 9:43 am

When all you have is a hammer, everything is a nail.

9 C June 24, 2014 at 12:11 pm

It really is the same old song. Low Productivity – High Illegal Immigration, coincidence?! Steve Sailer – SS, coincidence?!

10 Tom June 24, 2014 at 5:59 am

Still, even though I think he overly downplays resource misallocation, he’s pretty convincing that there has been a slowdown of IT-related productivity growth. And if you care about “potential output”, he’s got a relatively realistic take.

11 ummm June 24, 2014 at 6:04 am

I don’t see this as a big deal. Productivity is still growing. It’s lie when ppl says China’s economic growth is slow, but it’s still growing a lot, it’s just that the 2nd derivative is negative

12 Brian Donohue June 24, 2014 at 8:47 am

Oddly framed. Does a slowdown in productivity growth come from anywhere? Is it like a slowdown in the rate of manna falling from heaven?

13 collin June 24, 2014 at 9:08 am

How much of the productivity slowdown is simply the reverse of Mandel’s productivity gains of 10+ years ago, where good and services were produced in Chindia and brought to the US. These wages are increasing so the costs are increasing while companies are slowly on-shoring duties with stagnant US wages.


14 Donald Pretari June 24, 2014 at 10:37 am

” Or, panicked firms could have cut workers exceptionally fast and found temporary efficiency gains that reversed in the recovery.”

Good to see this possibility mentioned. This is my view based on Fisher that I argued against Prof. Mulligan in 08:


“I take the rising productivity and rising unemployment to show that many people are proactively and needlessly being laid off out of fear and aversion to risk. There is rising Productivity because the Demand is still the same, yet there are fewer workers. And there are fewer workers because the layoffs are due to the fear and aversion to risk.

The reason that the fear and aversion to risk mimics the behavior of an increase in the supply of workers, is because workers are being proactively and needlessly let go, not due to the fundamentals of supply and demand. It is a case of employers misreading and misdirecting the supply and demand now prevailing. After all, in order for supply and demand to work, it has to be perceived by acting human agents.

So, my thesis is that letting people go because you assume that demand will decrease, when demand doesn’t decrease, leads to a rise in productivity and a seeming rise in labor supply, which it is, only not because the workers don’t want to work, but because the employers have proactively and erroneously let them go, from misperceiving the demand.”

15 Dave Barnes June 24, 2014 at 11:25 am

Where did the productivity slowdown come from?
From economists writing about it.
Where did the productivity slowdown come from?
From Colorado legalizing dope.

16 Hasdrubal June 24, 2014 at 12:04 pm

Having been a network engineer in the late 90s through the early 2010s, I’m not surprised by this at all: IT provides massive productivity benefits but also suffers from pretty extreme diminishing marginal returns.

Give someone a spreadsheet and you multiply their productivity many times over. Give them a spreadsheet that does graphs and you’re still significantly improving their productivity, but not nearly as much as giving them the spreadsheet in the first place. Upgrade them from Excel 2007 to 2013? On average there might be some net effect in productivity, if you measure carefully enough, but it’s probably lost in the noise and possibly negative since the interface changed so much.

The same goes for networks: Put an office on Token Ring and allow them to share files and you get massive productivity improvements. Add a proper network file system and bump them up to Fast Ethernet and you get some more bonuses as actions that once took a minute or two now take half a minute or less. Bump them up from switched Fast Ethernet to Gigabit Ethernet and a Core i7 processor and they _might_ notice an improvement in the Youtube videos they watch in their spare time, but most of the actual productivity improvements will be in small fractions of a percent of the activity they spend the least amount of time on in a day.

Take it away and everything goes to hell, so you _have_ to keep investing in your IT infrastructure. But you’re mostly spending to keep the status quo and make some small incremental improvements. There are still some big gains to be had, but the vast majority of IT investment now is keeping up with the treadmill or making small improvements. The low hanging fruit has mostly been picked and the big improvements are fewer and farther between.

What was the trajectory of TFP growth in the decades after electricity became practical? I imagine it was similar.

17 The Kaigat Of Wands June 24, 2014 at 2:27 pm

Yes indeed – however the good news is that labour productivity is about to increase massively in this sector, unfortunately the bad news is that since it’ll come about through computers handling it all themselves there won’t be that many actual people being productive ………..

18 Jason Smith June 24, 2014 at 12:25 pm

The total factor productivity slowdown is likely monetary in origin:


19 Jake June 25, 2014 at 11:18 am

I’m surprised they don’t mention the biggest factor in productivity growth: wage levels. If labor’s expensive, it makes sense to spend a lot of money on improving productivity. When labor’s cheap there’s less need. So in times and places with good wage growth – like in the US from WWII through the 1970’s, and again in the 1990’s, you also see a lot of productivity growth. In times with stagnant wages, like the 1980’s and the 21st century so far, productivity doesn’t grow as fast.

Obviously there are other factors out there, but this seems to pretty much explain the macro level productivity trends in the US.

20 Darren June 25, 2014 at 9:29 pm

Following on Hasdrubal’s insightful comments above, I’d also be t that a significant and growing chunk of IT investment is I defensive in nature these days – things like cyber security and backup/archive (unless you’re the IRS) are huge and growing necessities that were largely ignored 10 years ago.

It was uncommon to have a single security expert on a large IT staff a decade ago (and no, the guy who installed mcafee on your PC wasn’t a security expert), but today there are commonly at least 2-3 security related teams (compliance / risk, network, desktop/endpoint, analysis, operations, etc).

In the scope of productivity c these very expensive people are overhead, as are their multi-million dollar budgets for hardware and software. Necessary? Absolutely, but they contribute nothing to productivity.

I’d be willing to bet that as IT has matured we’d be able to find other ‘hidden’ areas (these changes aren’t hidden to today’s CIO, but they aren’t readily apparent to a outsider) like this as well driving IT costs up over time. Combine that with being near the plateau phase of the ‘increased productivity’ s-curve and you have a recipe for a productivity slowdown.

21 Ryan July 12, 2014 at 8:38 am

Whatever your theory for the source of the slowdown, it must be reconciled with the large literature on reallocation and productivity:


In short, data from at least Manufacturing and Retail suggest that the majority of aggregate productivity growth is accounted for by reallocation–ie, entry/exit and other gross job flows. Entry alone accounts for around one third. Given what we know about declining rates of reallocation, some slowdown might not be surprising–at least in partial equilibrium. Technological innovation is necessary but not sufficient for aggregate productivity growth–technology can be improving rapidly but not show up in productivity if the market isn’t reallocating resources from unproductive firms to productive ones.

Productivity problems are reallocation problems.

Comments on this entry are closed.

Previous post:

Next post: