Robots confusing economists

Chatter on the econosphere has been abuzz on robots and income inequality recently, stirred into action by Paul Krugman’s NYT piece last week, and subsequent follow up.

We have Nick Rowe using this talk to support a general equilibrium approach to economics and make the mathematical case here.  Previously we have seen Google Chairman Eric Schmidt explain how robots will may result in income inequality.  A good summary of the whole blog-versation is here.

Debates like this, which used to take place in coded and mathematical language inside economic journals, are now raging online for all to see.  And it reveals the usual shallowness and confusion of economic thought on very critical matters – matters on which the common person assumes economists are experts.

Any economist who sees robots as anything more than another incremental change in the technology embedded in our capital stock is utterly confused. Similar debates raged a century ago when people called their contemporary robots ‘machines’.  But it is these very machines, the modern incarnations we might call robots, that are the methods by which we increase our productive capacity.

Detailed productivity analysis shows that it is primarily capital deepening, or the investment in a larger stock of capital, that leads to improved productivity.  New technology is the residual, or error term, after we account for our increased stock of capital – including machines and robots. Growth is a process of capital investment.

And as far as the arguments about income inequality arising from robots replacing skilled jobs, well that’s just nonsense.  Income inequality rises when productive capacity increases because the nation’s (or world’s) fixed assets of land and resources, are already privately owned, and the value of these assets is a mere reflection of the productive capacity of the economy that reside within. The more concentrated this ownership, the greater the rise in inequality when there is high growth in productive capacity, as a smaller group captures the now larger pie of economic rent being created.

That almost every economist misses this crucial concept of rent reveals a lot about the state of economic theory, and its relevance to the big policy questions in the real world.

Confusion amongst economists about technology stems from the type of optimal control problems typically used to model the economy, where capital and labour exhibit diminishing returns to scale and are compensated at their marginal productivity.  Because of the assumption of decreasing returns to scale,  these models show that at year infinity no growth can occur without something else happening. If the model has a constant term then a change in that constant is the only thing that can cause growth.  Let’s call that change “technology”.  Why? Because someone in the 1980s threw an idea out there and it stuck.

Why anyone thought we were at the end of time and that we need to invoke a change in the constant to explain economic growth is beyond me.

I mean think about it.  Say technology is actually the knowledge of how to make things.  Let’s send someone with lots of technology and human capital (the knowledge to make and use technology) to an uninhabited island, and call him Robinson Crusoe (a favourite story economists tell their students and children).  What on Earth is Crusoe going to do with all that “technology”? Clearly he would need to start at the basics, maybe carve a shovel from some timber with sharp rocks, then use that shovel to dig some trenches to divert and store water.  Amazingly, he produces capital (shovel) by means of capital (rocks) and the process of growth in his productive capacity ensues.

I don’t want to get too carried away with yet another rant about the relevance of mainstream economic theory, but this video sums it up nicely.

Tips, suggestions, comments and requests to [email protected] + follow me on Twitter @rumplestatskin

Comments

  1. A vision of an economy without labour

    “[T]hat interest must decrease as rent increases, we can plainly see if we eliminate wages. To do this, we must, to be sure, imagine a universe organized on totally different principles. Nevertheless, we may imagine what Carlyle would call a fool’s paradise, where the production of wealth went on without the aid of labor, and solely by the reproductive force of capital — where sheep bore ready-made clothing on their backs, cows presented butter and cheese, and oxen, when they got to the proper point of fatness, carved themselves into beefsteaks and roasting ribs; where houses grew from the seed, and a jackknife thrown upon the ground would take root and in due time bear a crop of assorted cutlery. Imagine certain capitalists transported, with their capital in appropriate forms, to such a place. Manifestly, they would get, as the return for their capital, the whole amount of wealth it produced only so long as none of its produce was demanded as rent. When rent arose, it would come out of the produce of capital, and as it increased, the return to the owners of capital must necessarily diminish. If we imagine the place where capital possessed this power of producing wealth without the aid of labor to be of limited extent, say an island, we shall see that as soon as capital had increased to the limit of the island to support it, the return to capital must fall to a trifle above its minimum of mere replacement, and the land owners would receive nearly the whole produce as rent, for the only alternative capitalists would have would be to throw their capital into the sea. Or, if we imagine such an island to be in communication with the rest of the world, the return to capital would settle at the rate of return in other places. Interest there would be neither higher nor lower than anywhere else. Rent would obtain the whole of the superior advantage, and the land of such an island would have a great value.”

    — Henry George, Progress and Poverty (1879), Book III, Chapter V, par.13.

    (And notice the assumption of mobile capital seeking maximum absolute advantage.)

    • It would be helpful if the national income accounts actually showed a ‘rent share of income’.

      An increase in rent share could explain flat real wages (decline in labour share) coupled with a low real interest rate (since roughly 2000). Which I think is what Nick Rowe is saying.

  2. I would like to believe it but I find it hard to work out in my mind how some technologies could not lead to higher unemployment. They may create extra jobs elsewhere but no way is the number or value equal over time.

    For example, retail positions at super markets. When self-checkouts pop-up everywhere, in addition to RFID checkouts, those jobs would be replaced by relatively cheap capital expenses. Some jobs that are being replaced are mainly done by young people because they are new to the workforce and have no skills. They can’t just train up to a different role. I can easily envision a world where the number of retail checkout people are diminished.

    What about when some company finally builds a nimble humanoid robot that could replace many cleaners, waiters and other low-levels jobs. Once they are mass produced (by robots), they’d be cheap and I wouldn’t be surprised to have them being able to perform basic maintenance on each other.

    I am not against this progress although I do believe it will cause problems and I am unsure how they can be addressed.

    • This is a common concern, and I share it. However, the critical factor here is the rate of adoption of this new technology/new capital. A large sudden change will cause a short-term loss of jobs.

      In my opinion, even a massive technology shock would rarely be able move employment to a noticeable degree in aggregate (at least within the bounds of the error of the statistics anyway).

      You actually point out exactly the process of progress – robots maintaining each other. Think of cars. Your mechanic has a bunch of robot-like power tools to maintain your robot-like car.

      So I guess my view is summed up as

      1. New technology requires new human management – whether in design, maintenance, distribution, coordination of contracts etc.

      2. People do take gains in production as leisure in aggregate and over the long term. The proportion of one’s life spend working is probably far less now than ever before (mostly from more years of study and more years of retirement).

      3. Low paying jobs will always exist, even if the are assisted by mechanisation. But the idea here is that fewer people will need to do them, and they can participate in higher level, more constructive jobs.

      • What higher level, more constructive jobs? Technology is getting to the point where people are being used as tools to assist automation, and the progress is accelerating. It is those higher level, constructive jobs that technology is targeting for replacement. Lawyers? Soon to be gone. Doctors? On the way out; the same for pilots, software engineers, experts of all sorts, and all but the highest levels of management. Scientists? Will not be that long. We have more than enough of bankers, venture capitalists, CEOs, and politicians. So again, What jobs will they be?

    • “I find it hard to work out in my mind how some technologies could not lead to higher unemployment.”

      The people replaced will be free to do new jobs that did not exist before. Think about going to the supermarket. Compare this to going to the market 200 years ago. There would have been nobody offering massages for $10 200+ years ago, because almost everyone had to work just to provide life’s basics. The industrial revolution changed this. Less people are now required to make necessities (e.g. food), so those of us that don’t provide basic necessities provide more luxury goods/services, which we see in our retail stores; e.g. entire stores just selling mobile phone accessories!

    • We can have more technology without adversely impacting employment. This is not to say it wouldn’t change the labour force and employment trends.

      Let’s say that we had the technology so that half of the labour required to grow corn or soybean (etc) was taken care of machines. Instead of needing 200 man-hours to plough, sow and harvest a field, it now takes 100 man-hours, plus a service of equipment.

      In this scenario the same crop volume is produced, just with less human capital.

      Because of this, if we were able to stabilise inflation and possibly impose or encourage some kind of state ownership of assets (or higher taxes etc) then we could, in theory, all work less hours to get the same output of goods/services to sustain our population.

      That’s more time with the family, and less time behind the computer or out in the field. This would also improve health (particularly mental health).

      Of course on a very practical level this wouldn’t work quite that smoothly, for reasons best attributable to human nature (particularly if in capitalist economy).

  3. reusachtigeMEMBER

    When the robots realise that we are of no use to them anymore they will extinguish us as they have no moral frame.

  4. No offence, but if your piece was written as clearly as Krugman’s I’d have a much better sense of what you are trying to say here.

    Which parts of Krugman’s thesis is wrong?

    • “…if your piece was written as clearly as Krugman’s”

      No Nobel prize for me then 😉

      “Which parts of Krugman’s thesis is wrong?”

      Actually, Krugman’s focus on monopoly power and the old capital vs labour debate is the right line of enquiry here. Though he needs to talk about property and labour, since capital is just stuff we produced in previous years that we still use. Property is a right, and is a monopoly on a piece of land, a right to fish, etc.

      His attention on robots as a likely contributing factor is confused. And the attention this idea has received from every economic blogger indicates they are focussed on the wrong questions.

      Even when Krugman cites Ricardo, his interpretation seems a little off. Ricardo talks of labour-displacing machinery being bad because if people are producing new machines they are not producing consumption goods. Thus large scale investments will come at a temporary cost (in the form of reduced output in other areas), even if the long run gain is high.

      What this has to do with robots I am not sure. I don’t see large swathes of US capital being diverted into robotics. However I do see a decline in the capital investment ratio to GDP, which suggests the opposite effect.

      Don’t get me wrong, Krugman made the most sensible comments in the whole debate – he usually does. But the idea that ‘technology caused inequality’ is not helpful.

      • No worry’s Rump’s,I’ll give you a Bell Prize reversed,just for the jinx on Algebra,cheers if you’d give me enough time to get my work-list done,read MB complete n tackle UE’s populations..acceleration I’ll tell’ya ,Id need two computers ..hey ‘Magine that..”Hey look,even wealthy countries can suffer creditworthiness-threatening financial crises”,could be back on screen,in harmony again n looking to others ,Financial Sense n more…n,that would be a wrap back to MB,UE,DE,while others just feel the Force,Today,n Cheers from me..n Merry Xmas All JR

  5. Regarding Prof. Rowe’s post.

    He starts this way:

    “Assume that all capital is robots, and robots are perfect substitutes for human workers.”

    For those not in the know, two productive inputs are perfect substitutes when one can replace the other at a constant rate without affecting the output. Say, one human worker is equal to X robots; you sack one human worker and put X robots in place and nobody is the wiser (indeed, for Prof. Rowe is one human for one robot without affecting the level of output)

    Then he adds this other assumption:

    “Let’s measure wages in terms of consumption goods. Because consumption is what people care about. Robots and humans earn the same wages.”

    So, if (by assumption, I should add) both human workers and robots can do exactly the same work and they earn exactly the same, is there any wonder in Prof. Rowe’s conclusions, namely

    “In this simple model, improving technology for producing robots has no effect whatsoever on wages.
    “Not at all nightmarish, is it?”

    What on earth makes robots and humans different in Prof. Rowe’s model?

    But there is a more elemental objection to the model: robots are supposed to be part of capital, and capital doesn’t earn wages; workers do.

    • One can look at it this way: Workers earn wages so that they can maintain and upgrade themselves to continue to present to work; robots have the same need, and if they are autonomous enough, they could be paid wages so that no other agent needs to be “paid” to optimize their productivity. The real objection to the model that it threats humans as equal to workers.

      • I agree with you in that the alternative interpretation you propose is possible, within the rules of the model.

        But I don’t think the wages thing is separate from the equal treatment. Quite the opposite.

        Consider the following Cobb-Douglas production function:

        y=%k*(X^%alpha)*(Y^%beta);

        From simple inspection of what’s written above, is there any reason to suspect that one of the two independent variables (X or Y) is labour and the other is capital?

        And even your alternative interpretation does not escape the contradiction: robots, whether more or less autonomous, are the capitalists’ property. It is the capitalists’ problem to provide for the depreciation of their property.

        Workers do not belong to the capitalists. It is their problem to provide for their own “depreciation” and for raising a new generation of workers.

        In a sense, your more autonomous robots are indeed human.

  6. What is the point of this? We already know that technology is beneficial because it increases productivity. We also know that there is always a use for labour in an economy even with new technologies/machines/robots replacing jobs. History has proven this numerous times.

    This reminds me of the anti-robot sentiment from the movie iRobot. Don’t be a luddite!

  7. Hmmm Robots replacing all Labor, what a beautiful concept.

    Automation and the construction of highly automated systems is a topic I can get truly passionate about.

    Most people hear the word robot and immediately think of clumsy a “lost in space” style robot protecting poor little Will Robertson. Cute idea but hardly an accurate model of today’s robotics.

    If wealth generation through automation is the aim of robotics then the most advanced robotics are to be found within the High Frequency Trading centers of the world. There have been stories of HFT systems generating over $1B profit in 1 year all for the cost of a few computers, a high speed link and an algorithm. Gosh $1B in profits from a robot trading with other robots on a microsecond timescale, about 1 million times quicker than humans could possibly operate. Phenomenal asset transfer occurred without a single brick being laid or joist heaved into place, no phones were created and no wonderful new drugs were developed, indeed in the real world practically nothing changed, yet wealth was transferred to the robots operator. The only necessary inputs are electricity, computers, energy and algorithms. WOW what a system, why would I aspire to create any other robot when this one works so perfectly?

    Lets step back and look at this concept again. Why do these HF trading robots generate any profits at all? Nothing in the material world has changes yet they extract a trading profit. Interesting!

    Consider another type of robot, a fully automated mine located in a distant deserted island country lets call it Pilli. The island was deserted but it contains desirable natural resources. The existence of this automated mine would have very little direct impact on the employment / under-employment balance in other neighboring countries, capital is invested and natural resources are transferred to other processing sites. The expansion or contraction of this mine would have almost no impact on regional employment, so how should the central bank manage their economy when capital demand is completely disconnected from labor demand?

    How about a construction robot with all the necessary licenses to do plumbing, brick laying, dry wall, electrical and roofing work. wow now that would be useful given construction costs in Australia. It is a bit of a dream today but hardly outside the scope of what could be achieved in 20 years. It is interesting to consider if such a robot would necessarily mean cheaper housing.

    • Hmmm Robots replacing all Labor, what a beautiful concept.

      Automation and the construction of highly automated systems is a topic I can get truly passionate about.

      And welcomed, you are a brilliant read.

      Most people hear the word robot and immediately think of clumsy a “lost in space” style robot protecting poor little Will Robertson. Cute idea but hardly an accurate model of today’s robotics.

      If wealth generation through automation is the aim of robotics then the most advanced robotics are to be found within the High Frequency Trading centers of the world. There have been stories of HFT systems generating over $1B profit in 1 year all for the cost of a few computers, a high speed link and an algorithm. Gosh $1B in profits from a robot trading with other robots on a microsecond timescale, about 1 million times quicker than humans could possibly operate. Phenomenal asset transfer occurred without a single brick being laid or joist heaved into place, no phones were created and no wonderful new drugs were developed, indeed in the real world practically nothing changed, yet wealth was transferred to the robots operator. The only necessary inputs are electricity, computers, energy and algorithms. WOW what a system, why would I aspire to create any other robot when this one works so perfectly?

      Hehe, exactly.

      Why would we be bothered to do anything in Australia when flipping houses made so much money too.

      Taxes channel incentives.

      Do we want HFT traders ticket clipping, or do we want product brought to market in ever increasing productivity gains.

      They aren’t the same robots.

      Lets step back and look at this concept again. Why do these HF trading robots generate any profits at all? Nothing in the material world has changes yet they extract a trading profit. Interesting!

      They aren’t doing anything extra, they are performing the same tasks as humans with increased efficiency and/or productivity.

      It speaks more amount how poor thought out banking remuneration is.

      Consider another type of robot, a fully automated mine located in a distant deserted island country lets call it Pilli. The island was deserted but it contains desirable natural resources. The existence of this automated mine would have very little direct impact on the employment / under-employment balance in other neighboring countries, capital is invested and natural resources are transferred to other processing sites. The expansion or contraction of this mine would have almost no impact on regional employment,

      Define the region?

      This mine would surely be the lowest cost mine in existence.

      It would supplant the highest cost mine, and all its employees.

      so how should the central bank manage their economy when capital demand is completely disconnected from labor demand?

      It can’t, and shouldn’t be expected to.

      The management of this isn’t a monetary policy issue.

      It’s like asking a plumber to manage faults with electrical wiring.

      How about a construction robot with all the necessary licenses to do plumbing, brick laying, dry wall, electrical and roofing work. wow now that would be useful given construction costs in Australia. It is a bit of a dream today but hardly outside the scope of what could be achieved in 20 years. It is interesting to consider if such a robot would necessarily mean cheaper housing.

      That is the way, it has always been the way.

      Why we always revert to radical shifts in tax, IR, welfare, etc.. is how to determine how this wealth creation is distributed.

      Capital (at a micro-economic levels) loves these sort of tech breakthroughs because it still views labour as an expense, not an asset (they both exist on the same side of the ledger).

      It wants to capture all the gains made from a diminished payroll.

      Game theory says all competitors will follow this path, it has to.

      Then it wonders where its customers have gone.

      Productivity gains have to be passed onto the consumers, and in most cases it has a tendency to do so.

      When the savings are passed on, it frees up discretionary spending, which creates new jobs.

      Jobs and weath creation is not trickle-down, its a vine crawling up.

      Business competes to supply product via either offering the product with the increasingly smaller margins or the increasing higher quality. Both which come about via productivity gains.

      Those that fail to keep up leave the market.

      To acheive outcomes that don’t see the wealth passed onto consumers are a result of an individual or individuals seeking power, its a rational and typical human response. But it is not enduring unless structures of power allow them to do so.

      Barriers to entry, capturing regulatory regimes, brute force even.