September 28, 2013
Tyler Cowen: Average Is Over

Tyler Cowen, author of The Great Stagnation, a book about the cause of slow economic growth and stagnant wages, has published his sequel: Average Is Over: Powering America Beyond the Age of the Great Stagnation. The title refers to how the average wage earner is disappearing as society increasingly fractures into two separate major classes, neither of which is centered in the middle. I hear Yeats: "Things fall apart; the center cannot hold".

For years I've been telling anyone one who'll listen that they better get higher level skills or else. So the theme of this book strikes a strong chord with me. Think you are firmly and safely planted in the middle of the middle class? Don't be complacent. You've got to move up or you most likely will move down (or perhaps you have already). Says Tyler:

This imbalance in technological growth will have some surprising implications. For instance, workers more and more will come to be classified into two categories. The key questions will be: Are you good at working with intelligent machines or not? Are your skills a complement to the skills of the computer, or is the computer doing better without you? Worst of all, are you competing against the computer? Are computers helping people in China and India compete against you?

If you and your skills are a complement to the computer, your wage and labor market prospects are likely to be cheery. If your skills do not complement the computer, you may want to address that mismatch. Ever more peopl are starting to fall on one side of the divide or the other. That's why average is over.

Don't feel too self assured just because you are getting paid to use a computer today. Lots of computer users who aren't doing intellectual heavy lifting are at risk of having their job eliminated. Look at people who work as cashiers. They essentially enter data into a computer. But increasingly people do their own store check-outs or they order online. Ditto bank tellers who compete with ATMs and online banking. Ask a software developer how hard it would be to automate your job out of existence. Then make career plans accordingly.

While Tyler envisions North American countries banding together to use robots in manufacturing I am skeptical of a major role in robotics for large political units that have large populations. It would make much more sense for capitalists and the most skilled computer workers to basically decamp to escape the taxing desires of the less skilled masses. The machine learning Ph.D.s and capital owners could set up shop in smaller countries such as Iceland where energy is cheap and the number of people that will want to lives as parasites of robotic factory output is much lower.

Outside of work I am struck by how much fewer people I interact with in order to carry out buying and getting services. I spend several times more money online than in person. Plus, I do not even use my own labor to make the computers do some of my bill paying and buying. My bills are mostly auto-paid. I even have standing orders for some dried fruits at a few month intervals. The grocery store is my biggest remaining contact with humans who work in retail. I look forward to the day when I can instruct my car to take itself in for service when it starts running rough or needs regular service.

One factor that slows down automation: lots of human environments are irregularly and complexly shaped. Consider homes for example. A plumber has to do some detective work to figure where a pipe might run and where it might be leaking or clogged. But imagine future humans built to a computer spec, their construction done by robots, and with lots of embedded sensors. These homes will be designed for easier automated repair and upgrade service.

Or look at food preparation. I expect we will see the spread of automated restaurants that serve foods whose prep is easiest to automate. Is pizza easier to automate than hamburgers or salads? Then we'll see pizza shops that serve more variations and other foods that are easiest to automate. People who need to save money will out of necessity choose services that require the least human involvement.

I am still reading Tyler's book and will do more posts about it as I progress.

Update: To anyone willing to state their current occupation in the comments I'll give you my guess on how long your job will last.

Share |      Randall Parker, 2013 September 28 11:23 PM 

Lono said at September 29, 2013 3:42 PM:

I can personally attest to the increasingly rapid dissolution of the middle class. A few days from now I was seriously facing the reality of living with my family in a "van down by the river" - but now - due to a highly paid IT position I was just offered - I can comfortably count on bringing in well over 100k again.

And - as a tall, fit, white, male - with no criminal record and a college degree - I have a considerable advantage over many who are looking for full-time employment in America. I would not want to be in the shoes of my old, mega-talented, guitarist who foolish accumulated three petty felonies right after High School.

I, for one, am now resolutely committed to making myself essential to the machine going forward. The great thing about programming computers is the ability of an employee to add value to people's lives without having to walk all over others in the process.

Bob Jenkins said at September 29, 2013 10:16 PM:

I'm surprised AIs have not yet taken over all thinking jobs.

Houses that weren't built to computer spec are easy enough to map out in a computer, if that should prove useful. Similarly, once computers build and repair and deconstruct physical things, for things that were built by hand, you can still instruct a computer about how to repair and deconstruct it. Objects don't even need the instructions built into themselves, if a computer can look up the instructions on the internet that's good enough.

ErisGuy said at September 30, 2013 5:27 AM:

So how much of this technologically-driven and how much politically-driven?

If the changes are caused by technology, they are inevitable with Green/Moslem suppression of Western technology.

If changes are caused by poor politics, a solution may be found.

An inability to supply more than basic needs to unemployed people unable to work should drive down the population, one way or another.

Russ in Texas said at September 30, 2013 6:55 AM:

I'm a reformed academic who's now an office manager/admin type in a sales office.

While the position's prestige is *vastly* lower, I consider it to be considerably more robust long-term (plus I like the private sector better than working inside a guild system). I think education has a bright future, but will look absolutely nothing like the sclerotic mess of today.

Blunt Instrument said at September 30, 2013 8:01 AM:

It seems that the coming mass automation will bifurcate productive society into two (broad) classifications: those who can generate ideas (via analysis of larger and more complex data streams) and craftsmen / craftswomen. Sadly, there will always be a place for rent seekers, apart from "productive" society.

FWIW: My current occupation is probably best described as "mechanical engineer".

bbartlog said at September 30, 2013 9:41 AM:

One issue with mass produced homes is that housing is a very important positional good. We already have the technology to manufacture fairly livable housing units for cheap: think of a larger and more modern version of the soviet apartment block. But people are willing to spend a large portion of their income differentiating themselves, and often that means non-standard modifications to their housing environment, or housing that's built in a an interesting and labor-intensive way from the beginning. Of course, right now one of the main reasons for expensive housing is that you're paying a huge premium just to exclude people with various social pathologies from living near you.
I work in software QA. Oddly enough I think this job may outlast many technically more demanding software development jobs, and if I manage to hold on longer than other software QA engineers before becoming obsolete it will be because I have better communication skills.

Anthony said at September 30, 2013 1:27 PM:

Mass production requires a mass-market. If we lose the mass-market, i.e. The middle-class, we will not have mass production. We need a guaranteed minimum income to save mass-markets.

Jack said at September 30, 2013 1:36 PM:

How about creative jobs? Advertising copywriter for example.

paul said at September 30, 2013 7:55 PM:

What is your prediction for a CMA or other accounting designations?

Randall Parker said at September 30, 2013 8:06 PM:

Blunt Instrument, Mechanical engineering. It has been heavily automated. I expect that trend to continue. Got some future. But more future if you can write software models of the physical system that a mechanical system will manipulate. A lot of physical engineering work has been automated. Draftsmen are mostly gone.


Automated systems will do more of the work of generating advertising creatives, even including some of the text. You've got to get seriously great at it and get in a tight loop with computer systems that feed back to you what does not work.


Manual testing gets automated when it starts to cost too much. I've personally proposed, architected, and was one of the implementers of automated testing systems that replaced manual testers.

Bob Jenkins,

We do not have AI yet. We have machine learning (ML) systems which are really valuable but are way short of AIs. ML systems need very smart people to design and iterate on them. ML experts have the longest lasting job prospects IMO.


The world has billions of people. But if a small number of people have lots of money then the mass production will come in the form of making all the bathroom fixtures for their 50 room summer homes and 100 room winter homes as well as parts for all their cars and boats and airplanes. I mean, the wealthy will not need the masses. They'll need their robotic factories, engineers, and natural resources. If you aren't an engineer or natural resource owner you still might be able to make it in security. But the competition will be stiff.

Ronald Brak said at September 30, 2013 11:30 PM:

Let's just say that Iceland builds lots of robotic factories and gains the ability to produce pretty much any manufactured item at a lower cost than anyone else. Canada might export resources to Iceland and get manufactured goods in return, but what's a country without much in the way of resources going to use to pay for Icelandic goods? They might have some gold reserves or works of art Iceland might accept for a while but obviously there are limits to this and these countries will end up without anything they can offer Iceland. Now a nation like South Korea isn't going to say, "Well played, Iceland, you win. We cede all manfacturing in the world to you and shall return to subsistance farming." No, as they have nothing to purchase manufactured goods from Iceland with they will manufacture stuff for themselves and will build their own robotic factories. Yes, their energy costs may be higher than in Iceland but since they can't buy stuff from Iceland it doesn't matter. And one of the first things they will probably do is produce a robotic factory that builds low cost generating capacity and bring their energy costs down anyway. A place with cheap energy can attract industries, particularly energy intensive things such as aluminium smelting, but it won't be able to monopolise manufacturing because plenty of places won't have anything to offer an Iceland that can manufacture everything it needs.

Anthony said at October 2, 2013 10:55 AM:

Randall Parker, my point is that new things will not be designed and manufactured for markets that are too small to create a profit. Consider the iPad, the iPad was not built for a market of only 1 million rich people. Apple must have a market of hundreds of millions of middle-class people who are able to purchase it or the iPad does not get built. Mass production requires mass markets to make massive profits! Read about economies of scale. When 99% of the disposable income is consolidated in the hands of 1% then you don't have a mass-market. One billionaire will only buy so many iPads. The billionaire, his family, and his billionaire friends are not a mass-market - they are not an economy of scale.

Randall Parker said at October 2, 2013 7:11 PM:


If the billionaires spend as much money as the masses would have spent then the factories will see as much total demand. It'll be different demand certainly. But the distribution of demand into a smaller or larger number of hands is orthogonal just how much total demand there is.

I agree that some products won't get built if only billionaires have money. On the other hand, whole classes of products will get built for billionaires that would not get built for the masses. Luxury automobiles, private jets that can range into the hundreds of millions of dollars, yachts that cost a few hundred million dollars each, mansions, large islands landscaped, rocket ships to ride into orbit. The list is long.

I also expect billionaires to spend big on life extension. How about paying people to grow extra organs in their bodies? Then paying lots of biomedical engineers to tend to all this organ growth.

Billionaires who group together to take over countries will pay most of the current residents to leave. Then they'll fund security and military forces.

Ronald Brak,

If a country doesn't have a lot of brains or a lot of natural resources its prospects will be bleak. A country that today can specialize in manual labor for low tech factories to make stuff to sell to rich countries (any country with people sewing buttons on shirts and pants for example) is going find it harder to make stuff that will let them get money to buy sophisticated goods. Once robots cost less than humans in all textile manufacturing then textiles close as the gateway for less developed countries to trade with the most developed countries.

ABC said at October 2, 2013 9:49 PM:

Total demand for mass products will go down because the wealthy don't buy mass products. They consume luxury goods and positional goods - goods that are defined by the very fact that they're not mass products, that they're rare or unobtainable by most people, that they don't have substitute goods for, etc. In such a market, mass product producers can't survive. Only smaller scale luxury and positional good producers who produce low numbers of high priced products survive. In fact for such goods, higher prices and greater exclusivity increases demand from the wealthy. The suppliers aren't going to produce large numbers of products in this market when they're going to go unpurchased and when even if they were purchased they would lower the perceived value of the product and the price it can command from the wealthy.

Ronald Brak said at October 3, 2013 2:14 AM:

A country with low literacy rates, low economic development, and limited resources will certainly be in trouble if the demand for unskilled labour decreases. But robots and automation pushing down the price of labour is not all bad for them as it will lower the cost of capital goods they require for economic development, so robots will give with one metallic claw and take with the other. Countries that have made a good start towards development may also be okay. Looking at India, while its exports may seem very large at 18% of GDP, India still has a huge informal economy that isn't included in GDP figures and perhaps less than half of exports are light manufacturing which is the area most at risk to robots. (India is the 10th largest agricultural exporter in the world and labour is already a very small part of the cost of goods produced by heavy manufacturig.) If foreign demand for Indian manufactures dries up they can concentrate their manufacturing on meeting India's own huge internal demand. Note that I'm not saying there is no cause for concern. Export industries can impose discipline on industry and have benefits beyond the foreign exchange they bring in. But the world is a different place than it was a generation or two ago and it remains to be seen how things will actually work out. And as for China, well, they have already shown ample ability to mobilize capital and use it in the service of industry. (Basically they steal it from the savings of workers and loan it to industrialists at around zero percent interest. Karl Marx would be so proud.) China is quite capable of building vast numbers of robot factories.

What does seem clear is that countries with low energy costs will have an advantage as robots will decrease labour costs and thus increase the portion of the cost of manufactured goods that is energy, but it will also lower the cost of building generating capacity which will make energy cheaper and somewhat limit that advantage.

Nick G said at October 3, 2013 1:08 PM:


"If the billionaires spend as much money as the masses would have spent"

That's a big if. Billionaires tend to spend a much smaller percentage of their income. That's part of our current economic problem: rising income inequality reduces consumption "demand".

Nick G said at October 3, 2013 1:55 PM:


Automation will only reduced energy costs. With the primary exception of oil, which has a scarcity premium, all energy costs can be pretty much boiled down to labor, eventually.

IOW, energy is a manufactured product, and falling manufacturing costs will reduce energy costs.

Ronald Brak said at October 3, 2013 8:13 PM:

Nick G, we only get energy through people doing stuff, so I see what you're getting at, but that's not how the term labour is generally used. In this case I'm refering to paying people to do stuff. Businesses have capital costs, labour costs, and energy costs. If I am going to start a sausage making business I am going to need a sausage making machine which is a capital cost and I'll have to either borrow money from the bank or buy it outright. If I borrow money I have to pay interest on the loan and if I buy it outright I have to forgo the money I could have made from investing my money elsewhere. Either way it costs me. I could reduce my capital cost by making sausages by hand which will reduce the capital cost required to make each sausage but dramatically increase the labour cost. If I pay someone to make sausages for me, either by hand or using the sausage making machine, then that's a labour cost. I will need power to run the sausage making machine, lights, refrigeration, etc, so I'll have to pay for electricity. Most of my electricity bill goes towards paying the capital cost of transmission infrastructure, wind turbines, thermal coal plants, coal mining equipment, and so on and a small fraction goes towards paying for the labour of the miners, technicians, engineers, software designers, accountants, etc. So energy cost is really just a combination of capital and labour costs, but from the point of view of my sausage making business I have to pay for energy and so I have an energy cost. I can do things to reduce the my energy costs. For example I could buy a more efficient sausage making machine, install LED lights, or install rooftop solar. From my point of view the rooftop solar system is a capital cost as I don't care what the installer paid for labour, I just care what I had to pay for the system and its a capital cost that reduces my energy cost. As long as it reduces my energy costs by more than what the capital costs it's a worthwhile investment for me.

Randall Parker said at October 3, 2013 9:11 PM:


Of course a higher IQ country like South Korea will fare better. Their own engineers will be able to write software, design chips, and develop sensors. They'll be able to do genetic engineering too.

But there is going to be a big shift in trade. The natural resource owners and robotic factory developers and builders will trade with each other. The competition for natural resources will be won by those which can offer the most to the natural resource owners.

Randall Parker said at October 3, 2013 9:13 PM:


On the one hand more automation is coming to accounting. So lower skilled and less demanding accounting work will dry up. Accounting will get even more automated. The trend of exchanging more financial data automatically will continue.

On the other hand, managing the money is very important and accounting jobs have been growing. You'll need to operate at a different level to survive in accounting. If you know how to create offshore corps to legally dodge taxes you could have a bright future. Ditto if you know how to restructure a business's money flows to avoid regulatory mandates. Ditto if you know how to reduce risks.

WCN said at October 5, 2013 3:11 PM:


Agree with much of your argument, but I think there will be a very bright future for those that have skills - and I mean REAL skills - working with their hands (e.g. plumbing, masonry, general construction, etc) for the foreseeable future. Plumbing consists of much more than simply the toilet, sink and even the pipes in your home or office building. There is an entire complex network of water supply, sewer and storm drainage piping that runs throughout the cities and towns of our country. Plumbing (the entire network and not just our toilets) is one of the unsung things that makes us so much different from third world nations (I know...I've worked there)

I can say the same about road builders, and commercial and residential construction. People will always need roads, homes and places of commerce and I don't see very much automation - other than better tools and construction equipment - that will replace much of that workforce.

FYI, I am a Municipal Utilities Contractor

Randall Parker said at October 5, 2013 3:24 PM:


Automation is coming to guiding building construction. Robots are coming to construction sites.

Check out a 20 foot tower built by flying robots.

New construction is a lot easier to automate than maintenance because with new construction everything is open and the sequencing of tasks could be completely controlled. Plus, the choice of materials and design can be made to fit what sorts of materials and layouts robots could best handle.

WCN said at October 5, 2013 5:09 PM:


I see the robots augmenting the manpower just as excavators and bulldozers do now. Of course, as I alluded to above, less unskilled manpower will be needed as the machines improve (similar to the way bulldozers eliminated the need for more men with shovels, but not all), but I don't see our industry being completely eliminated for at least another century or so - at least.

A few robots building a simple brick tower is a far cry from building a complex 500,000 sq ft convention center or an expressway. Men will certainly use such machines to help make the job faster and easier, but it will not completely eliminate the men (at least for the near future). Some unskilled men will be eliminated, but the skilled guys will be around.

Secondly, most new construction is not as rote as you seem to imply. There are almost always unforeseen conditions (electric line running in some place unexpected, bad soil conditions, etc) that require changes to the plans in the middle of construction. In other words, building a simple house off the grid would be relatively simple for a robot, but constructing a new convention center or mixed use building requires an extraordinary amount of coordination that at this point requires lots of skilled men.

In short, I don't foresee an army of robots on a construction site (which don't forget includes schools, fire stations, hospitals, convention centers, office buildings, sports stadiums and all types of underground utilities as well) for a very, very long time.

A real game changer in plumbing would be some form of waterless disposal of waste and cleaning. That would save billions of gallons of water and eliminate the industry as we know it. I don't see robots doing that.

Randall Parker said at October 5, 2013 5:40 PM:


The extraordinary coordination: I understand about the unforeseeable. But consider what Chinese construction companies have accomplished: 30 story building built in 15 days and 6 flood building in 24 hours. A Chinese company that does this does most of the work in factories that make modules with wiring, piping, and other pieces already installed. That factory setting is key to enabling greater use of robotics.

WCN said at October 5, 2013 8:09 PM:

Yes, but

a) I still see many workers in those videos putting the buildings together
b) The assembly in the factories should count toward the construction time. The Chinese seem to only be counting the time of assembly.
c) One major obstacle to buildings being constructing so quickly here in the US are the state and local building codes, the unions where applicable, and OSHA. In the normal construction process, the various building inspectors check for leaks and defects, do soil density compaction tests, and other tests in each step of the process (underground, rough-in/top-out, and final construction). Then of course all of the wiring, piping and such have to be installed by licensed tradesmen (or under their direct supervision).

I do see your larger point though, and I do agree that better construction machines will eliminate more unskilled jobs, but I don't think that automation will eliminate or revolutionize our industry for a very long time. The real issue, as I see it, is that many young people do not have these skills and indeed tend to look down their noses at this type of work. Meanwhile the pay gets better for these skills.

Randall Parker said at October 5, 2013 9:29 PM:


I bet the Chinese factories are running roughly in parallel with the final on-site assembly process. Otherwise they'd pile up too many pieces with no place to put them.

I also bet they get a big labor savings out of this approach. Plus, they get a big time savings. Saving time saves capital. Think about a skyscraper that takes, say, 18 months to construct. At the 9th month probably half the money has been spent and none of the spent money is providing a return on capital invested. Fast construction reduces the amount of idle capital.

US regulatory obstacles: Understood. I think the Chinese approach could achieve higher quality because the factories could build up skills in their workers that would get used across jobs for different buildings and internal processes at the factories could be a lot more consistent than what will happen at each building site. A state and city that would rework their regulatory process to fit the Chinese model could get a big competitive advantage for companies with cheaper office space.

I expect the large housing construction companies (e.g. Pulte Group, D.R. Horton, Lennar, NVR, Toll Brothers) to embrace robots and prefabricated pieces. They've got the management depth. They've got yearly revenues over $1 billion each (over $4 billion for Pulte).

Check out Behrokh Khoshnevis's Contour Crafting approach to building construction. He likens it to 3-D printing. He is a USC engineering prof. I can see why this would be a hard sell in heavily regulated cities. But industrial companies that want to construct buildings in rural areas (e.g. at mines, in forests for wood processing) and farms could use this sort of tech without the need to fight unions and big city regulatory agencies. Ditto military bases that are pretty much exempt from local building regulations.

Pay getting better for the skills: there is a macro trend where wage premiums for higher skill levels keep going up. How poor do kids have to get before they get a clue? The labor market is sending large and growing signals that you'd better get lots of skills or else.

I work with the best at what I do (software development) and I know the best are just so much better than the rest that the differences are staggeringly big. Plus, I see us automating so many tasks and we keep just rolling along doing it.

Nick G said at October 5, 2013 10:24 PM:


The capital costs you're talking about are reducible to labor as well. The sausage making machine is manufactured by people: people deliver the machine, they assemble the machine from parts, the parts are assembled from smaller parts, and the final individual parts are manufactured from raw materials which are...mined and delivered by people.

Energy costs are reducible to labor as well. Coal is dug up by people, the coal is moved by people (with vehicles manufactured by people), it's burned by people (in plants built by people), and the power is delivered by power lines installed and maintained by people.

Solar electricity is manufactured in the same way, starting with sunshine and labor.

That's why energy costs will tend, in the longterm, to decline, despite the recent temporary scarcity premium (aka "rent") caused by the transition from oil to electricity.

Randall Parker said at October 5, 2013 10:54 PM:

Nick, I think you left out a step when you stated:

That's why energy costs will tend, in the longterm, to decline

Can you elaborate? You think labor will become more productive and therefore it'll become cheaper to deliver energy?

WCN said at October 6, 2013 5:41 AM:


Yes I can see prefabrication becoming more automated at a certain point in the construction process, but there are a lot of steps in the construction process that the average person doesn't see or pay any attention. Long before a single nail is driven in the first home in a Toll Brothers subdivision, people like me come into the 1,000 acre plot full of trees that is nothing more than a small forest or former farm land or what have you and perform lots of essential (and ultimately unseen) work. We remove trees, do soil testing and replace rocks, boulders and unsuitable soil, and dig trenches and place the piping for the water supply, fire hydrants, sewer and storm drainage. We also run underground electrical lines for the street lamps, and grade the land and develop the streets. The entire process can takes months depending on how large the subdivision is going to be, and other factors such as nature trails, and lakes will be in place. Then even after the streets and underground utilities are done, many homes have basements, which require digging and the removal of soil and the concrete must be poured in place. Other future homes are on lots that require a lot of grading and leveling. Even homes without basements, have underground footings that are dug and filled with concrete. Oh, and the underground section of the plumbing is placed (and tested) before the concrete is poured on the slab. From there, the process can admittedly be more automated - if they can figure out a way to test the various systems before occupancy (keeping in mind that plumbing inspectors like to see each and every joint of piping to make sure it is free of leaks when filled with water) Then there are highway, bridge and road widening projects that just can not be automated - at least not very much at this point.

Over-regulation not withstanding, the building codes really protect us from faulty building, fire hazards, plumbing cross connections (a massive danger) and other threats that I am not sure that the Chinese gov't is so concerned about. And though the military bases and Corps of Engineers are exempt from local regulations, they still have their own inspectors, building processes, and safety codes that can be quite fastidious. The Feds would probably be one of the last to come on board with such a thing.

To be clear, I am not a person that is for keeping 100 jobs around just for the sake of jobs if a single machine can do it more efficiently. But I am just demonstrating just how much skilled men will still be needed for a very long time.

I see this contouring process catching on more in developing nations (with large multi-national conglomerates such as Bechtel) than here.

So, I am just curious as to how long you think that my occupation will last.

Ronald Brak said at October 6, 2013 8:08 AM:

Nick, it's easier to write about labour cost, capital cost, energy cost, and raw material cost than it is to write about labour cost, other people's labour embodied in the form of a sausage making machine cost, other people's labour embodied in the form of electrical current cost, and other people's labour embodied in the butchered flesh of slaves animals and barley and pig's intestines cost.

Nick G said at October 6, 2013 12:48 PM:


Yes, I left out a step: manufacturing is relentless about increasing labor productivity (as well as reducing other costs). You can assume that any manufactured process will reduce it's labor input by a minimum of 5% per year (for a mature product and process) and much more if the operation is relatively new.

That's why manufacturing employment is *much* lower in the US than it was decades ago, across all industries, despite output being substantially higher. That's a large part of why wind & solar costs are falling, and why battery costs generally fall over time.

We often loosely call this process "automation", which makes it sound related to computers, but it's been going on since the beginning of the Industrial Revolution. It includes time study, line balancing, operations analysis, standardization, improved parts design, work simplification, numerical control, robotics, six sigma, logistics optimization, etc., etc., etc. There are a trillion ways to reduce the time required to make something, and that process will continue until that time is so small it's no longer worth the cost of measuring it. And then we'll reduce the cost of measurement.

It's why the Dutch threw their wood shoes (sabots) into the millworks, and created "sabotage"...


The point is, the cost of both capital equipment and energy will continue to fall, so that their relative costs will probably stay roughly the same.

Nick G said at October 6, 2013 12:55 PM:

Randall, Ron,

I suspect that economists tend to distinguish between capital manufactured goods and consumer manufactured goods because of their differential impacts on the economy. Capital goods are investments/savings which increase output and productivity, while consumer goods are just...consumed. Economies which accumulate capital goods are on a better path than those which consume their "seed grain".

Nick G said at October 6, 2013 3:37 PM:

As a practical matter, the same labor productivity trend applies to farming, mining and pretty much all "goods" production.

Services are a little harder to improve, but the major problem, I think, is measuring labor productivity for services. Attorneys, for instance, can research and write briefs much more quickly now, and online court filing will likely improve things. But, I'm not sure how the BLS can capture those improvements, some of which will show up as improved quality, rather than quantity.

How does the BLS measure the impact of the quality improvement from IOS6 to IOS7?? The reduction of time spend opening the iPhone? The increase in communication made possible by ubiquitous communication?

Nick G said at October 6, 2013 3:40 PM:

Let me clarify that last: many people check their phone 30 times a day. If you save 3 seconds each time by using a fingerprint reader, that's 1.5 minutes per day, or perhaps .3% of a workday. .3% spread over the overall economy is $50B!

Gimlet said at October 7, 2013 6:22 PM:

Attorney (most of my work is contract drafting and negotiation).

Nick G said at October 7, 2013 7:36 PM:


Contracts can be packaged in software, edited by clients or paralegals, and negotiated by the principals. Surely you're really being paid for your professional judgment, rather than your word-processing?

Greed said at October 9, 2013 12:52 AM:

There will be a guaranteed minimum income in the future promoted by the billionaires to keep them safe from assassination. I see a positive correlation concerning levels of inequality and number of assassinations. To be honest, I am surprised Rockefeller, Carnegie, J.P. morgan, and Vanderbilt were never targeted for assassination by some poor unemployed bastard with nothing to live for.

Randall Parker said at October 9, 2013 10:58 PM:

Nick G,

Relentless increase in labor productivity? But how? Much of it was done by utilizing more non-muscle energy to do the work. Most notably, more work was done by energy generated by burning coal, oil, and natural gas. Work today is done by a combination of human labor and machine labor. Much of that machine labor is powered by fossil fuels. If the fossil fuels become more difficult to extract then unless other energy sources become cheaper the effect is to reduce the productivity of human laborers. They end up needing to do more of the labor themselves.

If the cost of solar power declines to the point where solar power can replace fossil fuels then I can see how we can keep raising labor productivity. But right now we are very heavily dependent on fossil fuels for getting large amounts of work done.


As for how long your job will last: Your problem is actually bigger than that. Why? Anyone who can do your job but who now works in a job that will get automated before your job gets automated will want to compete with you at your job. So you don't just need a job that will not get automated. You also need a job that won't attract a lot of future unemployed people to it that could do at least part of what you do.

My advice: Look for ways to help yourself a lot by taking smaller steps toward greater productivity. Ask yourself what you could get better at in the next month that would be an additional layer of skills that would help to set you apart. Then ask what you could do in the next month after that and the following month and so on. Think of it like compound interest. How can you accumulate skills that will enable you to take on jobs that are more skills building so that you get into a virtuous cycle that pulls you further above those who will get laid off.

Nick G said at October 11, 2013 10:16 AM:

Much of it was done by utilizing more non-muscle energy to do the work.

I think that way of looking at it will mislead you: ingenuity is far more important than the particular source of energy. For instance, improvements in the reaper ( and cotton gin ( improved agricultural productivity enormously without a requirement of fossil fuels for operation.

Modern factories relentlessly reduce labor inputs without significant additional energy inputs. That's why manufacturing employment in the US has dropped by more than 50% in the last 40 years. Note that oil consumption in the US has dropped since 1979, while manufacturing output has risen by 50%, and GDP has risen by 150%.

They end up needing to do more of the labor themselves.

The value of extrasomatic energy is far too high for that to happen. Wind power costs in the range of $.05-.20 per kWh, and solar is $.10-30. Human labor only produces about 100 watts, so even at 30 cents per kWh, wind/solar can replace simple human energy labor for 3 cents per hour.

One risk of renewables is that a *modestly* higher percentage of the labor force would be devoted to producing and managing energy. That's not a big risk. A much larger risk is that we won't move away from FF fast enough to avert large climate change.

The greatest cost is the diversion of R&D and engineering talent away from other challenges, like improving human health and longevity.

Randall Parker said at October 12, 2013 12:01 PM:

Nick G,

You are distinguishing between increasing the value of a given amount of energy expenditure versus increasing the total amount of energy expended. An important distinction for sure. But our living standard increase isn't just due to devices like the cotton gin that gets more work done per unit of energy used. We also use the farm tractor (which likely uses more energy per amount of field plowed or planted) and fertilizer and other chemicals for planting and harvesting. We've got the truck and train that moves the harvest on to textile mills and then to stores and then to home. We've got many more times the energy getting used in the entire supply chain than is used to run the cotton gin or its modern equivalent.

My point is that energy used per person has gone up by orders of magnitude in spite of all the ingenuity. Our living standards are the result of both higher energy efficiency and much higher energy usage.

Per capita oil consumption: US living standards have stagnated for decades. If we really were able to detach our living standards from per capita energy consumption then we ought to have far higher living standards. We increased efficiency of some devices. But that energy efficiency gain has been eaten by the need to use more energy to get more incremental output from mines with declining mineral concentrations, from using more marginal lands for farming, and other areas of diminishing returns.

Nick G said at October 12, 2013 2:36 PM:


Our living standards are the result of both higher energy efficiency and much higher energy usage.

This is true. But, the way it's framed is misleading: thinking of the cotton gin as increasing energy efficiency focuses on energy, when we really want to focus on human work, measured in hours, not joules: the gin allowed the production of much more cotton with much less human time.

Yes, if we were to lose our extrasomatic "energy slaves" we'd be in big trouble. But, there's no realistic risk of that. Wind, solar, nuclear are all feasible, scalable, and far, far cheaper than human somatic energy. Again, 30 cents per kWh is a very conservative maximum for the cost of renewable energy, and that's very, very affordable for all of our needs (especially with efficiency improvements that have already been shown to be practical and cost effective even at much lower energy prices).

Energy is essential, but so are a lot of things, and ultimately ingenuity is far more powerful. After all, why didn't the Romans use coal? Because they hadn't figure out how to do it (they hadn't built up enough accumulated tech knowledge), not because it wasn't there. They were surrounded by energy: coal, oil, wind, solar, nuclear, but they didn't know how to use it.

Now, we do.

US living standards have stagnated for decades.

Well, no, they haven't. Median hourly wages have stagnated, but real GDP has gone up by 150%, and real GDP per capita rose by 60% from 1979 ( . The big problem: the rich are getting richer, and no one else is (IOW, deteriorating equity of income distribution). The 2nd problem: livings standards include a lot more than average income: service productivity and quality isn't measured well by GDP. Think about what you said about ordering things online. Add in better quality in a variety of consumer goods and services, including portable music, smart phones, etc., etc.

using more marginal lands for farming

Farm productivity per acre continues to improve. This is ingenuity, or "tech", or at work. I'm not as familiar with mining costs, but I believe that changes in tech, like chemical leaching, have kept mining costs from rising. Ultimately, of course, things will be recycled. That might increase costs for a while, but a new cycle of cost reduction would begin and ultimately reduce costs to the current level and then lower.

Randall Parker said at October 13, 2013 8:16 PM:

Nick G,

We aren't going to fall all the way back to a pre-industrial level of productivity. But right now high energy costs are a throttle on economic growth.

Electric power costs: electricity is still a far more expensive and inconvenient way to power cars.

The St. Louis Fed graph: the amount of total debt per unit of economic growth ballooned over that time span. Also, the US trade deficit went way up. We just went thru a decade of no net improvement. That's huge. Private debt is almost back to 2007 levels. Government debt is much higher. The United States is becoming a tougher place to live for well over half the population. As for the people who make more: lots of them are serving international markets and using lots of brain power to do it. They are basically avoiding the impact of the nation's overall decline.

Engineer-Poet said at October 14, 2013 12:42 PM:

I find electricity to be a cheap (after the battery is purchased) and very convenient way to power my car.  I just plug it in when I get home and (mostly) forget it.

The major issues come when driving outside of round-trip-on-battery range.  At that point the options devolve to (a) find a charger and park wherever it is (limiting the options somewhat), or (b) fall back to gasoline.  If we had a much larger penetration of plug-in vehicles, chargers would be ubiquitous and (b) would occur only on long trips.

Nick G said at October 14, 2013 2:35 PM:


I'm not sure we're communicating well, here. Let's try to look at a basic question: the productive capability of the US economy. I see an economy which can produce more than ever, and whose capability continues to grow. I also see an economy which is mismanaged, with difficulties in income distribution, debt and taxation management, and aggregate demand, so that output is below the potential (aka, the "output gap"). Does that make sense?

Ok, on to details:

We aren't going to fall all the way back to a pre-industrial level of productivity.

I don't see any reason for any fall in productivity. Productivity continued to increase during the Great Recession, as it did during the Depression. Heck, increasing productivity probably contributed to both of those events, as it increased unemployment and reduced aggregate demand.

right now high energy costs are a throttle on economic growth.

Energy costs aren't especially high in the US, even with moderately expensive oil. Coal, natural gas and electricity are all cheap. By historic standards, oil is only a bit more expensive than usual.

I agree that energy imports are a drag on growth for importing countries. On the other hand, they help growth in exporting countries. All in all, the net impact isn't very large.

In the US, you'd expect to see that impact in the form of inflation, that would force the Fed to raise rates. But, there's no sign of inflation, and the Fed isn't raising rates at all. The real problem: a lack of demand, caused in part by reductions in government spending, and in part by excessive income to the wealthy, who don't spend as much.

electricity is still a far more expensive and inconvenient way to power cars.

As E-P notes, electricity is far more convenient than pumping gas. An Extended Range EV is far more convenient than a pure gas car.

The Chevy Volt is cheaper over it's lifetime than the average new car: it onliy costs about $5k more, and will cost about $20k less to fuel. Even if we compare it to a much inferior gas car of about the same size, like a Chevy Cruze, it's still at least as cheap to own over the whole lifecycle.

the amount of total debt per unit of economic growth ballooned over that time span.

Yes, debt is becoming less efficient as a way to increase growth. Blame the wealthy, who are forcing the government to borrow from them, rather than tax them with properly progressive taxation.

For instance, WWII was financed by both debt and taxation. On the other hand, the Iraq war was financed entirely through debt.

The US economy (and the world economy) is underproducing, due to inadequate aggregate demand. The easy fix is more government demand to replace inadequate private demand: build energy infrastructure, for instance. But no - instead the right wing is forcing government to cut spending, keeping us in slow growth, and cut taxes, forcing up debt levels.

the US trade deficit went way up.

Well, no. The current US trade deficit is at about 2.7% of GDP, which is well below recent levels, largely due to dropping oil imports.

WCN said at October 15, 2013 10:10 AM:

Thanks Randall,

We can certainly agree on that part. One of the factors that insulate me personally is that I am a contractor (owner) and not a 'worker' and ironically I see the best way to increase my company's productivity in continuing to invest in the new technologies in our field.

Post a comment
Name (not anon or anonymous):
Email Address:
Remember info?

Go Read More Posts On FuturePundit
Site Traffic Info
The contents of this site are copyright