Technological Unemployment: Not Just for the US

by Andrew McAfee on September 10, 2012

In a talk I attended the other night, Larry Summers mentioned that manufacturing employment in China peaked in 1996. I found this hard to believe, so did a little Googling. Lesson #1: don’t doubt Larry Summers’s command of the facts.

Lesson #2: companies all over the world are automating rapidly. Even though hourly manufacturing labor costs in China are only 4% of those in the US, it’s still attractive for Chinese factories to replace people with technology over time. This allows them to turn out more stuff with fewer people. How much more, and how many fewer? Take a look:

Chinese manufacturing and employment, 1990-2008

China’s manufacturing output was over 70% greater in 2008 than it was in 1996. Over the same period, manufacturing employment in the country declined by more than 25%.

Obviously, this is not because of outsourcing (companies outsource to China, not from China). It’s because technology is now so cheap, useful, and universally available that when more and more Chinese companies upgrade an old factory or build a new one, they don’t fill it up with cheap labor. They fill it up with hardware and software, just like we do in the US.

An earlier post showed what’s been happening to US manufacturing and employment over time: output goes up, while the number of jobs goes down. Thanks to a tip from Summers, I now see that exactly the same thing’s been happening in China, the world’s other manufacturing powerhouse.

Anyone want to suggest another plausible explanation (beside automation and the resulting productivity growth) for what’s going on in these two countries simultaneously? Anyone predict that either of these trends is going to reverse itself in either country? Anyone think anything different is going to happen as automation spreads to more and more industries?

If so, I’d love to hear about it.

  • Anonymous

    Manufacturing employment is interesting but a bit on the marginal side from the point of view of the whole society. More interesting is the development of these issues in the context of services. For instance,
    we might imagine the hospitality industries maximizing the efficiency of their employees by placing them in a central location somewhere and having them use remote controlled carts or something like that for
    the actual floor or field work — to wait on customers. I can vaguely imagine sales people interacting with customers remotely using Ava-like platforms. The common thread is using technology to distribute labor, and particularly cheap labor, more efficiently. That would be plenty disruptive all on its own. Full automation can come later.

  • Nick Bloom

    Great post (forwarded on from Erik). One question is any idea if any of this is definitional? That is that jobs that used to be called manufacturing – say cleaning, canteens or providing day-care for SOEs – are now defined as non-manufacturing? If, for example, going from Government to non-Government owned meant that all the auxiliary stuff factories used to do changed from being defined as “manufacturing” to “non-manufacturing” employment, that could explain some of this.

    But frankly I doubt that’s much of the story, if any, and you are right this is a fantastic data point and example of the impact of technology

    Cheers

    Nick

  • http://andrewmcafee.org/blog Andrew McAfee

    Nick, thanks for weighing in. You ask a good question, especially because the BLS report I (and apparently Summers) relied on for the Chinese manufacturing employment stats is careful to point out that much data from that country is spotty, unreliable, and pretty opaque. So the phenomenon you describe could absolutely be taking place.

    I’m with you, though. I doubt that even if it’s taking place it’s a big enough deal to sway the overall story, which is one of rampant automation and productivity growth.

  • Jim Harvey

    Since 9/11 and, especially, after Katrina, I have been telling all that would listen that each American needs to have a skill that cannot be outsourced or, given this article, automated! In difficult times, I do not want to be standing on a freeway off ramp with a sign that says, “Will do technical support for food.”

  • jodi

    Something doesn’t add up, 300million Chinese have moved from the country to the city, what jobs did they take if not in manufacturing?

  • David Masterson

    I think you’re missing the big picture — productivity gains vs. the need for further productivity. As the productivity of the individual worker improves, unemployment goes up as there is less need for workers to achieve the same amount of productivity. As the scope of the market changes over time, the need for productivity may go up or down which, in turn, may increase or decrease the number of jobs available. Hopefully, the need for further productivity outstrips the increased productiivity of the individual worker and, thus, employment goes up. In the past, changes in the scope of the market happened through innovation that resulted in more jobs because only people could cope with the change — our support systems were not adaptable to the innovation (at least, initially). However, we are fast approaching the singularity in time when our support systems will be adaptable to (most of) the changes from the innovation and, so, the need to raise employment to cope will be (largely) eliminated. At that point, we will see (near) 100% unemployment and it is likely to happen (nearly) overnight. Remember, also, that in the pool of employees, most are going to be laborers (however you define that) and not innovators. The ratio of innovators to laborers has always been heavily skewed toward labor. When the singularity occurs and labor is (largely) eliminated, we’re going to have a *LOT* of bored people who are not qualified to do much of anything in the new market. There is only so much that education and retraining could do to affect this.

    This is the pessimistic view — now give me the optimistic one. The predictions of the innovations that will make this pessimistic view come true are 25 – 100 years.

  • Anonymous

    There are jobs that are quite clearly not likely to be automated, if for no other reason that we (the customers) don’t want them to be automated, and these are jobs that flourish in strong economies. Services. I come from Norway, where unemployment is currently a non-issue, even if we’re damn good at automating manufacturing. We have hairdressers, spa personnel, and so on. If you have the money to pay for a hair-do, do you want a droid to do it? Child care. Child care penetration in Norway is very high, and so is the percentage of parents working. Do we want droids to take care of our children? Even if the doctor uses automated diagnosis tools – do we want the person telling us the results be a droid?

    At some point in the future I am sure that just the novelty of having people rather than droids do the work will allow for a lot of jobs.

    And if most of us don’t have to work to live well, imagine how much time we would have to spend on books and movies and other entertainment! Would we want that to be written by droids?

    There is an issue with skill level, but I think that those with less skill will still find adequate work in services where persona is just as important as the job done.

    @google-f2a4b57538fad0d1028d29c71101cdc3:disqus I believe in the Singularity as much as you do, but labor isn’t only manufacturing. Even if unemployment drops (I’m not convinced, there is as of yet no proof that this has happened due to automation anywhere), there will still be a need for services, artificial or not, and with the ability to pay for it, this will grow rather than shrink from what it is today. I may still be proven wrong, but I don’t think that’ll happen in my lifetime (unless we’ll find us self living forever).

  • David Masterson

    No, labor isn’t only manufacturing — it is anything that doesn’t require much thought which separates it from the work of innovation. In the next 50 years, our support systems will probably evolve where most labor can be setup to be done in an automated fashion and, thus, largely eliminate the need for laborers. There will still be a need for innovation, but the ratio of (need for innovation) to (need for labor) has always been heavily skewed toward labor. Eliminate the need for laborers and I really doubt that the need for innovation will be able to pick up the slack to provide the needed jobs — in other words, I see a HUGE unemployment problem in our future without much chance of correction. In the past, shifts in innovation have resulted in shifts in the need for labor as only people could adapt to the new environment. In the future, our automated systems will evolve to be largely adaptable to the shifts created by the innovation and, so, the shifts in the need for labor will be much smaller (and probably evolve to being smaller and smaller). We will need a new economy!

  • David Masterson

    Already happening to some degree. I work at a large corporation that has a large number of buildings where employees work. To reduce the need to have an admin sitting in each building to greet visitors, the visitors are now greeted with a TV set that automatically connects to the central admin who interacts with the visitors and helps direct them where they need to go. It won’t be long before the central admin sends a robot cart to the visitor and instructs the visitor to follow the cart to where the visitor needs to go. Also, with many companies replacing phone operators with AI programs to help a customer get what they need without human interaction, it shouldn’t be long before this combines with robots to provide mobile service. When FaceTime becomes more prevalent, the question and answer AI programs may give way to “show me the problem”. In other words, I don’t believe that the service industry is a long term replacement for the loss of jobs in the manufacturing industry.
    Does anyone have statistics on the true unemployment of the world? (ie. people of reasonable work age who are not working.) I suspect that it is is in the 20% or more range and, so, feel that we need a HUGE amount of jobs into the future and there is not much that will achieve that.

  • Bob Ritter

    Technological is slavery without a victim. Technology is freeing us. Humans have always wanted servants to build their dreams. Now dreams can be built with fewer limitations. It adds up to greater human achievement without the drudgery. Union leaders dont see it that way, because they don’t focus on the value of the vision of the dreamers.

  • Mirza Duratovic

    It depends on the incentives. A pair of hair clippers costs $20 in Australia and a hair cut at the hair dressers costs $15. Lets assume you get your hair cut every two months for the next 3 years. The hair dressers would cost you $270 whilst the hair cilppers DIY would cost $20 and the clippers will last 3 years. This is what I decided to do, simple shave cut. That way I also don’t have to use shampoo. Which saves me another $150 over three years.

    In summery hair dresser = $270+$150 =$420 over three years
    Hair clippers = $20 over three years

    In order for the hair dressers to compete they would have to offer a hair cut service for $20 over three years. I don’t see how any of their workers would be able to pay rent with such low prices. If on the other hand your happy with paying $420 over three years that your choice but understand that there will be a economic shift over to people such as myself who’s cost of living is cearly lower then yours

  • Anonymous

    @mirzaduratovic: So you save a lot of money on not cutting your hair – what do you use it for instead? You won’t earn more money than me just because you don’t cut your hair at a hair dresser. There are no bonus points in this world for persons that are more “efficient”, except that you may be more flexible in terms of serving your future economical needs. Also, you have to admit that most people don’t want little or no hair :) Where I live, it is even terribly cold to do so.

  • Anonymous

    David Masterson I’m not really denying this possibility, but there are no current evidence that this is happening, nor that it will happen in the future. You cannot take the situation in the US, which clearly has a lot of other things going on than just the automation of jobs, and apply it to the world. You are assuming that the US is the perfect system, the blueprint that the rest of the world is created from. It isn’t. And neither can you (or Mr McAfee) wantonly disregard inequality or expensive education or other issues and just decide that “OH MY GOD – THE ROBOTS TAKE OUR JOBS!”, especially when there are ample evidence from around the world that that could indeed be reasons for what we see.

    So yes, at some point in the future, we may become job less due to automation, but it has never happened in the past, is not happening now, and so if it happens in the future it is so unprecedented that it must be considered a black swan. It may happen, in which case we clearly must deal with it, but what I’m saying is that innovation (which is still accelerating) has always been followed by economic opportunities and new economies, and I think that will continue into the foreseeable future.

    (And about here I discovered that I thought you responded to a more elaborate comment that I made much more recently than the above, and for some reason I thought you commented in the last day. If this reply makes little sense, you may see if you find the more recent one … sorry :)

  • Warren Harding

    Our governments should spend more of our tax-dollars on something we want. I’m for robotics that are owned by all of the citizens of a country. Lot’s of people want to cheer every time a robot puts someone out of work. Fully automated robotics factories, with self replicating robotic arms. Highly automated renewable energy, windmills or underwater water mills. Highly automated steel production. Highly automated chip manufacturing, and Linux. I’ve seen some automated building manufacturing companies starting up as well. Other prerequisite products can eventually be manufactured as well. All source code and blueprints have to be fully owned with rights to an infinite amount of use. All owned by the citizens of the country concerned. Small factories at first, with all of the bugs worked out, so that it largely builds itself in the end. It should be affordable, I’m an economic conservative. Eventually the complex can produce consumer goods besides steel, energy, chips, buildings, and robotics. Charities and the open source community can help as well. SymbolicComputation.com supports liberal licensing agreements of source code and blueprints, to allow infinite replication without cost(one time fee models).

  • http://www.facebook.com/people/Paul-Reverse/100001663413335 Paul Reverse

    “Death by Technology” by Dan Thomas – He is the Truth!

  • Mirza Duratovic

    How can that be true? Every company on the planet function under this model. By not being efficient you are giving stuff away for free. In the case mentioned above. The money will be invested which in turn will generate profit for me. Next time you go buy something on credit just before you swipe the card take a minute and think about me. Since I function under a 0% Debt hence 0% Interest system. By taking out credit you are effectively paying me. Then after you have made that purchase you are going to go to your job and your going to have to work in order to earn the money to pay the interest. So theoretically your not really working just for your employer. You are also working for me, the bank or for whoever gave you the credit card in the first place.

Previous post:

Next post: