Rise of the Machines: The Robot Economy of the Future
by Benjamin Studebaker
My attention has been drawn to a rather interesting phenomenon by Paul Krugman–that of a gradual shift in the distribution of wealth from labour to capital. As a percentage of the economic total, workers are earning less and less over time, and more and more of our output is landing in the hands of people who own capital–the land, the buildings, the tools, the machines that make things tick. This has interesting implications, and, if I might be permitted to speculate today, those implications may demand changes in how we view what we produce.
First, how do we know that the distribution is shifting in favour of capital? Here’s the data:
Around 1980 (why do nasty things have a habit of beginning around 1980?) we start to see a decline in labour’s percentage of GDP, from nearly 60% and falling. With the exception of a blip during the bubbles of the late nineties, that trend has continued, accelerating during the last decade. This brings us to two questions:
- Why is this happening?
- What does it mean going forward?
The answer to the first question, as Krugman argues, may well be capital biased technology. A technology that is biased toward capital shifts income from labour to capital. How does it do this?
Let’s say that we have a box factory that manufactures 10,000 boxes per week. Initially, the boxes are put together by hand with the aid of basic tools–saws, nails, hammers, and the like. In order to get these boxes completed on time, the workers complete via an assembly line. Imagine, if you will, that there are 100 workers required to achieve this (I know nothing of box-making, the number is picked out of thin air–it is merely an example). So what we have are relatively low capital costs for the various basic tools, and relatively high labour costs, to employ the workers who wield the tools. Now let’s say we introduce a capital-biased technology. Say someone invents box-making robots. With box-making robots, many more boxes can be manufactured in a given period of time with much lower labour costs. Say, for instance, that in order to make the 10,000 boxes per week with robots, only 10 workers are required to oversee and maintain the robots. The cost of these robots is likely greater than the costs of the basic tools, but the labour costs are lower. Unless the demand for boxes increases dramatically, we will see the same number of boxes being made with less labour, so more of the company’s income will go not to the individual box-makers, but to the person who owns the robots, the capitalist.
As we automate more and more industries and services, larger and larger proportions of our wealth go not to human employees, but to the owners of machines and robots. This explains why we’re seeing a decline in labour’s share of GDP, but what of the implications?
As it turns out, the Congressional Budget Office makes an interesting assumption in its future projections–it assumes that the decline in labour’s share of GDP is a blip rather than a long-term trend that will continue going forward. This means that much of our data about the future of the American economy is based on an assumption that may well be false.
How does this matter? As it stands, payroll taxes (the taxes that fund Social Security) are capped beyond a certain income threshold–around $100,000. This means that any income over this threshold is not taxed for the purposes of collecting funding for Social Security. As wealth redistributes in favour of capital and against labour, more of the wealth is going to people who earn over the threshold, which robs our entitlement programmes of much of their funding. Estimates vary and the future is difficult to know, but it’s estimated that removing this cap on payroll taxes would extend Social Security’s ability to be self-sustaining for 75 years.
Even more importantly, the shift in favour of capital and against labour creates, from a Keynesian point of view, a long term demand problem. If most of your consumers are receiving a smaller part of GDP with which to consume, it becomes more and more difficult for those consumers to sustain the economy’s present level of production. As we saw with the recent crisis, consumers are having to rely increasingly on borrowing to make up for the income they do not receive. This creates inherent economic instability and is in part responsible for our weak recovery–consumers have quite a bit of debt to pay off before they can resume their spending:
It is no coincidence that the increase in household debt begins around the same time that labour’s share in GDP begins declining. The private sector has left consumers (and workers, for they are broadly the same people) in a bind. It requires that they supply the demand to keep the businesses running, but refuses to pay them the sort of wages that facilitate doing this in a sustainable way.
Going forward, we may find it very difficult to sustain economic growth, particularly if the trend away from wages and labour continues. Labour’s share shrank by around 3 percentage points in the last decade, from around 57.5% to 54.5%. Should this continue, indeed, should this accelerate, as robots get better and get more widely implemented, we will be faced with an interesting choice. As more and more workers are made redundant by machines, we can choose one of two paths:
- Evolved Status Quo–the owners of the robots become fabulously wealthy without doing much of anything, unemployment becomes permanently very high as very few workers are needed to produce the supply demand will support, and what demand there is comes almost solely from redistribution via the state. The government taxes the robot owners to provide welfare to the unemployed masses, the masses buy what products they can from the owners. We end up with a very slow or non-existent rate of economic growth, lots of poverty, and all of the various problems associated with welfare dependency (crime, alienation, unrest, mental illness, violence, and so on).
- Left Wing Alternative–the robots become the common property of all, everyone receives a universal basic income funded by the work of the robots, work becomes a thing of the past for most people who instead become free to pursue their passions and interests, the few people who do work receive larger shares to provide incentive, the economy grows as fast as we can augment the production of the robots (as supply and demand become, essentially, one and the same), and we broadly move in what can only be described as left wing utopian direction.
I submit to you that the second society is by far the more preferable one in which to live. I further speculate that the first kind of society would eventually fall prey to a revolution in favour of the second, as the overwhelming majority of people would be unemployed and have much to gain from the latter scenario. In order to do it, however, we will have to overcome our primal fixation on claiming right of property over something merely because we had something to do with its manufacture or purchase. Do not mistake my meaning–the person or people who make and buy these robots in the first place deserve some reward for their initiative, but it cannot be a reward in perpetuity, willed by law to their offspring, as the current laws would make it. It would need to be temporary, one time only, and end upon their deaths. An extremely high rate of inheritance tax would do the job–if you are involved in automating a part of the economy that previously required lots of manual labour, you should receive a very large reward for doing so, but it should return to society upon your expiration so that the fruits of your work, the output of the automatons, work for the benefit of the workers they replaced rather than against them.
Our automation must benefit the workers it replaces. It must not infantilise them into welfare addicts. Paradoxically, to avoid a state in which welfare is near-boundless, we must create a state quite a bit more socialist in nature than the one we presently have. In this bizarre paradoxical future, to avoid the nightmares of the political right, we may implement policies the political left has, at present, refused even to acknowledge to be its own–namely, very high inheritance taxes, and communal ownership of automated industries. Should this sound like simple communism to you, note that the only sectors of the economy in which this could reasonably be done are the sectors where incentive does not matter. A robot does not care how much it is paid. It does not care who benefits from its output. It can be put to work either for an individual or for everyone; it does not care either way. It will still perform its function–robots are diligent that way. So why not take these robots, assuming they do arrive, as they seem to be slowly but surely arriving, and use them to emancipate us from labour rather than shackle us to welfare?
An interesting take. Does it consider the lowered barriers of entry and reduced costs and risks associated with compounding technological advancements? Robot technology is being taught in some high schools, and you can build your own at home.
As a software developer, I’m sure that side of this has to factor into the equation; someone has to be able to translate human behavior into algorithms and signals for these robots to understand.
In so far as lowered barriers go, it depends on the scale and sophistication of these robots when they are developed, I suppose. The sort of machines that automate the automotive industry at present are very big and very costly. If automation progressed along that line, I doubt that barriers to entry would drop very far, but if that changes and robots get simpler and smaller, it may very well be possible for barriers of entry in some industries to fall, industries that do not have other large costs (the car industry will likely always have high barriers to entry because of the inherent expense and specialised skills involved in designing the cars to be made, for instance).
As I understand it, getting the robots to be able to understand human behaviour is the big obstacle at the moment to having say, robot doctors or robot mechanics. The complexity of the data sets (say, giving a robot knowledge of the possible symptoms of every known disease and having it diagnose patients based on that data) does not seem like a substantial challenge in and of itself. It’s the interaction (and perhaps the trust and liability if the robot makes an error) that keep those parts of the economy human-run at the moment. That and the financial interest of current employees in preventing robots from making them redundant, of course.
Is this the book that is referenced by McAfee & Brynjolfsson, in Race Against the Machine? I ran across the title, Economy of the Future in an Amazon Review of their book. The other book was James Moody’s book which had to do with “waves” of economic cycles.
This is my own work–I was not aware of and have not read that particular tome.
Yes, I know its your own work, what I wanted to know was, is it the book they referenced.
Nothing in this piece is in reference to that book.
Hi there, do you have a machine answering your emails? You wrote a book. And after you wrote a book, they wrote a book. And because theyre good academics, they want their reader to know who else is writing on the topic, so they make reference to other people’s books, but they get the title wrong, or incomplete. So I am asking you again, is it your book that they referenced, that they want people to know about, because they admire your book, its is yours?
I haven’t written any books by that title.
[…] we now can manufacture many goods more cheaply with fewer workers–we can automate. But the robot economy has never been met with universal glee–those made unemployed by it have raged against it. […]
[…] the wealth to themselves and leave the rest of the population in poverty (as on Earth). There are debates as to whether this will lead to the return of socialism or an increase in government redistribution, […]
[…] the robot economy? Regular readers might. A little over a year ago, I was considering the extensive effects on both […]