Where Digitization Is Failing to Deliver
University of Chicago economist Chad Syverson explains technology’s curious lack of impact on worker productivity.
Topics
Frontiers
It has become a truism that the pace of work is faster than ever, as digital technologies speed up communication and operational processes in a story of unending progress. But increased speed has not translated into increased rates of productivity growth. Since 2004, growth rates have slowed not just in the US but across the world. Commentators have questioned the way growth rates are measured, suggesting that the data fail to capture the effects of new products. But in a working paper published by the National Bureau of Economic Research in February, Chad Syverson, J. Baum Harris Professor of Economics at the University of Chicago’s Booth School of Business, argued that the slowdown in the rate of productivity growth is real. MIT Sloan Management Review asked Syverson what the implications are, and why the benefits of new technologies are not as straightforward as we think.
What do the data say about the pace of growth over the past decade?
The official productivity measures have shown that there’s been a slowdown in growth rates since about 2004. It seems pretty clear it started before the financial crisis and the Great Recession. Productivity growth since 2004 on average has been half of what it was for the decade previous. (Productivity is output per worker hour. In the end, it’s an efficiency measure: How good are you at converting inputs into outputs.)
If you believe the trend is real, the question is, why has productivity slowed? While you have all this talk about how much technology is marching us forward, the numbers just don’t show it. In fact, we’re in a period of unusually slow productivity growth.
There are a couple of explanations. One is that we really aren’t coming up with innovations at the rate we were before, at least for now. Why? Maybe because we’ve wrung out the easy gains from IT and other innovations, and now we’re in a period when it’s getting harder to find productivity-enhancing applications.
Another potential explanation is that the frontier is moving: The possibilities are there, but they aren’t being picked up, they aren’t diffusing as fast as they used to. So while the leading edge might be moving out at the same rate it always has, the average company in the economy is not keeping up.
One historical example of this is the struggle of the Detroit 3 — General Motors, Ford, and Chrysler — to adopt modern (“lean”) manufacturing methods. They were way behind the Japanese initially, and then thanks to a combination of early denial, followed by grudging reluctance, and eventual struggles with implementation, it took them decades to catch on. GM even had a joint venture with Toyota, called NUMMI, which worked well and demonstrated how to do it, yet the Detroit 3 still couldn’t translate that success to their other plants for years. There’s some nice new evidence that that is part of the story right now, that best practices don’t seem to diffuse or percolate through to mainstream business as quickly as they used to.
Get Updates on Innovative Strategy
The latest insights on strategy and execution in the workplace, delivered to your inbox once a month.
Please enter a valid email address
Thank you for signing up
When you talk about productivity, are you referring to both office-based and manufacturing work?
Yes, both. The official productivity measures cover about 75% of the economy. They are based on non-farm private business, so they leave out government and non-profits, and they exclude farms, which aren’t that big of a part of the economy. But they pick up pretty much everything that’s going on, both high-tech companies and low-tech, traditional companies. The rate of productivity growth isn’t the same across different sectors of the economy, but the slowdown is pretty ubiquitous.
And this isn’t just in the US. I reviewed data on about 30 countries — mostly western industrialized countries — for which the OECD has tracked productivity growth since at least the mid-nineties, and you see the slowdown in almost every one — with similar timing too.
It’s a big shift, and it’s important because productivity growth drives the growth in standard of living. If you have a slowdown that lasts a decade or two, even if it’s only a percentage point a year, it really adds up. In the US, it has fallen by about 1.5% a year since 2004. It might not seem a lot, but you add that up over 11 years, and we are now missing about $3 trillion per year — $3 trillion with a ‘t’ — $3 trillion dollars of output that we would have reaped had productivity growth not slowed down.
Why do you think we have this assumption that things are moving forward so quickly if in fact the growth rate is slowing down?
That’s a great question. Silicon Valley gets a lot of mindshare, but in the end, it isn’t that big of a chunk of the economy. There’s all this other stuff that goes on that’s less directly connected to it, and it just seems like for all the progress that we seem to be hearing about, either one, it’s not spreading, or two, not to say it’s illusory but it’s just not as big of a deal as you might think it is. Maybe it’s cliché at this point, but Peter Thiel’s “We wanted flying cars, instead we got 140 characters” line articulates this sort of what’s-the-big-deal-isn’t-this-disappointing attitude.
How do rates of productivity compare going back further in time?
In the US, post-war, between 1947 to 1948 and 1973, productivity growth was very brisk, around 2.8% per year. Around 1974, it slowed down, dropped by half and stayed sluggish for 20 years. That was a big puzzle at the time, a paradox. Bob Solow famously said, “You can see the computer age everywhere but in the productivity statistics.”
Then in 1995 it sped back up to roughly where it had been in that post-war period — in fact maybe a tiny bit faster — and kept at that pace for a decade until 2004, when it slowed again. Today, we remain in the second productivity slowdown of the IT era.
For historical context, we could go back to 1890. The analogue to IT around the turn of the twentieth century was portable power: electric motors, internal combustion engines and electricity. 1890 to 1915, as new technologies were taking root, was a period of slow productivity growth. But then in 1915 — just as in 1995 — the rate of growth sped up for a decade, only to slow again in 1925, just like it slowed after 2004. The difference is that by the time we got to the mid-1930s, productivity growth had increased again, whereas we remain mired in slowed productivity growth today.
We haven’t hit that second spurt for the IT era. I hope we will for all the reasons we talked about, but we don’t see it showing up yet. Past history has taught us that technology’s impact can come in waves; that a set of technologies doesn’t have to have just one big kick and then it’s over.
How might the current slowdown reflect decisions taken by company management? Does it suggest that company leaders haven’t been able to integrate or harness the potential of new technologies?
Both of those possibilities — the slowdown in innovation rate and the percolation (or diffusion) rate, how fast a technology spreads — could be very easily thought to be related to management practices, maybe not solely, but in part. We know management practices matter; they’re related to productivity, and if for some reason managers are not as able or not as willing to implement best practices as they were before, that would affect the percolation rate.
In terms of innovation, if managers are thinking too short-term or not being smart about how they allocate their research dollars, then that can also lead to a slower innovation rate — by which I mean the probability that a novel, actionable idea will arise from an effort to create something new. There are different ways one could misallocate research dollars: by going after things that might be achievable but are trivial, or going after things that are important but infeasible. So both the frontier of innovation can slow down and the diffusion rate can slow down based on whether managers are doing what they ought to be doing or not.
Could there be other explanations?
It’s not implausible that when new technologies come along, there’s a period where you’re spending so much effort trying to figure out how to use them that although the old way of doing things could still have gotten gains, you’ve decided to spend some cost figuring the new thing out. You accept a temporary slowdown in return for (the hope of) faster growth in the long-run.
And I do think that that’s one reasonable explanation for the slowdown at the beginning of the IT era, 1974 to 1995, and for that matter could have explained the slowdown at the beginning of the portable power era back at the turn of the century. If enough new technologies come along, there are adjustment costs, there are implementation issues, which can lead to a temporary slowdown while everything is put in place.
And I think that’s one reason why you can get the waves that we were talking about. The first wave is just the replacement. You take the new technology and you replace what you were doing before directly with the new technology. The second wave comes from recognizing that you can do other, truly new things now with this technology — not just replacing the activities you did before but tapping multiple fresh possibilities that interact with this technology in exponential directions.
For us, that hasn’t happened yet?
No, it hasn’t happened. Take inventory, for example. A lot of people think IT helped inventory practices, and it did. Basically what we did was, we took pencil and paper and we replaced it with a computer, but we’re still doing inventory in the same way. It’s just we’re doing it more efficiently now that we have a computer.
So you could imagine once you’re tracking goods electronically, it’s not just that you do inventory differently, you do all sorts of things differently. And while we’ve gotten little bits of that, maybe, there’s a whole bunch of possibilities we haven’t figured out.
Comments (2)
Tinko Stoyanov
GUS CAWLEY