Are robots and aging demographics self-cancelling problems?

Dean Baker says yes:

This is one where a baseball bat might be necessary. If you are concerned that a falling ratio of workers to retirees is going to make us poor then you are not concerned that excessive productivity growth will leave tens of millions without jobs. Let’s try that again. If you are concerned that a falling ratio of workers to retirees is going to make us poor then you are not concerned that excessive productivity growth will leave tens of millions without jobs.

It is possible for too much productivity growth to be a problem, if the gains are not broadly shared. It is also possible for too little productivity growth to be a problem as a growing population of retirees imposes increasing demands on the economy. But, it is not possible for both to simultaneously be problems.

That is missing the point, as there is too much talk of “productivity growth” per se and not enough of either distribution or political economy.  If robots concentrate wealth in the hands of IP owners, wages for many workers might fall or remain stagnant.  That is a problem.

Similarly, if robots concentrate wealth in the hands of IP owners, it may be hard to drum up the tax revenue to support a higher dependency ratio.  The wealthy may produce a blocking political coalition or capital simply may be harder to tax for mobility, accountancy, and Laffer curve-like reasons.  There is then a problem with the dependency ratio.

We then have both problems, no contradiction.

Note that we can get out of at least one half of this mess if the robots are especially good at taking care of old people.  That seems unlikely to me, at least in earlier stages of robot development, but of course it is not impossible.  In reality, many old people would fare somewhat better if our economy were somewhat more like that of Mexico, namely with cheaper land and cheaper servants.  You could imagine robots lowering wages by say ten percent, yet still labor wouldn’t be cheap enough for most old people to afford much more in the way of servants.


It all depends on whether more young or more old people will get killed in the final war between humans and robots.
Based on the experience of previous wars, it's going to be the young.
=> This robot thing is far more terrible for demographics than previously thought.

In Mexico, do the servants have servants?

No, not in Mexico. But in India they do. And servant's servants have they own servants, etc., all the way down. That's why there are so many people over there.

I'm for robots replacing tenured faculty.

not sure that robots will do it first. but I can see colleges and universities hiring a few folks from low cost countries to do it remotely. could even have one 'professor' teach thousands of students in any US college or university. just thing of the cost savings might have have a few teachers aides (usually graduate students). could even do that with economics.

What the writer is calling the problem of "aging demographics" is really a tax revenue problem -- not enough people paying into Social Security to support those receiving it. Robots do nothing to alleviate that, since they aren't paid wages that are subject to Social Security taxes.

I'd suggest a new tax on the employment of robots, but I can't imagine how it could be enforced.

The real problem, of course, is that governments have made it entirely too rewarding not to work, and people have responded to those incentives.

RE: "I’d suggest a new tax on the employment of robots, but I can’t imagine how it could be enforced." -

That could be enforced more easily than on humans. If a robot doesn't pay its taxes, you can confiscate it and auction it off. Or use it as a mine-clearing robot. - Far easier than dealing with human tax dodgers.

No, the "real problem" is not that governments have made welfare too rewarding (you give a link to an article on the UK welfare system). Instead governments like the UK`s have implemented Milton Friedman`s doctrine that unemployment must not be allowed to fall below its "natural rate", which is why during the last boom the ruling class became very nervous about falling unemployment - a Financial Times editorial published on September 11, 1998 told its readers that unemployment would have to rise, "perhaps by 500,000"! Three months earlier the Bank of England`s Monetary Policy Committee (MPC) had raised interest rates to 7.5% after concluding that "it was probable that unemployment would have to rise to hit the inflation target".
Higher interest rates are used to increase unemployment during booms by reducing consumption and new investment.
The economists on the MPC not only want "enough" unemployment, they also want the "right type” of unemployment. Six months after Chancellor Gordon Brown created the MPC, in its December 1997 meeting the Committee asked itself "did short-term unemployment exert more downward pressure on earnings than long-term unemployment?". They concluded that "short-term unemployment was more important", on the grounds that "when the proportion of long-term jobless was high.....workers would probably realise that they could not be replaced so easily, and hence that their bargaining strength was higher".

Further reading:
The Bank of England is licenced to rig the labour market

Whenever someone suggests that living on the government dole is easy and plush, I have to assume they have never been anywhere close to poverty (college years supported on Mom and Dad's dime do not count), nor had anyone close to them in that situation. Sure, welfare is (barely) enough to keep body and soul together-- and would anyone not either insane or evil desire a civilization where people starved in the streets?-- but no one has anything but the most utterly threadbare and perilous existence on the dole.
Now, if you would like to discuss instituting world requirements with public jobs provided for those able-bodied but "unemployables" the private sector has dumped like yesterday's cat litter, by all means I am happy to listen.

@Andreas Moser: The reasons I expect it to be hard to tax robots are that one could fairly easily conceal having them; and that in some cases it's hard to count them. (Take one of the automated car factories in Japan as an example: does each individual machine on its assembly line count separately, or is the whole thing one big robot?) It would be like trying to tax computers.

@AndrewC: Do you have a link to Milton Friedman saying that unemployment should be kept above a minimum? That is a crazy and inhuman idea and I'd be astounded if he advocated it.

@JonF: I did not suggest that living on the dole is "easy and plush", only that the system has severe disincentives for getting off the dole once you are on. (It is hard not to believe that these were created deliberately by the welfare agencies, lest they lose their own jobs by solving the problem of poverty as we're supposedly paying them to do.)

As for "unemployables" "the private sector has dumped," I don't feel there should be any such category of people, even if it means adopting Germany's unemployment system (which puts recipients on a sequential waiting list and finds jobs for them). This is one area where I cannot be a pure libertarian -- it is not only inhumane but totally unreasonable to expect people to behave themselves if they're put in the position of having to steal to get their basic needs met.

Actually, it's not missing the point, it is Tyler not reading the whole piece he is citing. I did in fact write later in this piece:
"It is possible for too much productivity growth to be a problem, if the gains are not broadly shared."
I have also written hundreds (thousands?) of times about how distribution is in fact a huge problem. And, I have argued endlessly that our obsession with demographics is a distraction from the problem of distribution.

So Tyler strikes out badly here. If he wants to make an argument that distribution can be a huge problem even in the context of rapid productivity growth, he is attacking the wrong person.

It is silly and perhaps indicative for you to think I am "attacking" a "person." My post doesn't attack you at all. (Put away the baseball bat, by the way.) Your key sentence is "it is not possible for both to simultaneously be problems." -- which is in fact wrong and for the reasons I point out. Nothing in your comment overturns that.

Well, selective quoting is quite a problem, especially since you literally left off the entire opening from Baker. There is an expression about not first attributing something to malice when another explanation can be found (such as reading comprehension, perhaps?) -

'Paul Krugman had an interesting column today calling attention to the rise in the profit share of income. The point is that we seem to be seeing rapid improvements in productivity growth (he cites the progress in the development of robots) that are drastically reducing the demand for labor. Yet all the gains from these improvements seem to be going to owners of capital as the labor share of output has been falling sharply.

The distributional issue raised by Krugman is extremely important, both for workers who are not seeing gains in living standards, and also for the economy as a whole, since a continual upward redistribution of income will lead to stagnation as a result of inadequate demand.'

So, where in the opening is 'That is missing the point, as there is too much talk of “productivity growth” per se and not enough of either distribution or political economy.' not acknowledged in the actual text? Or is 'The distributional issue raised' at the opening of the second paragraph, or 'continual upward redistribution of income will lead to stagnation as a result of inadequate demand' at its end not the proper formulations the general director of the Mercatus Center (sorry, can't provide a link to here due to what appeared to be automatic filtering of links in the past) prefers when referring to the work of the co-director of the Center for Economic and Policy Research ( )?

And why isn't your writing 'If robots concentrate wealth in the hands of IP owners, wages for many workers might fall or remain stagnant. That is a problem.' simply not a restatement of what the linked test opens with, to a major extent?

Or did someone simply miss the point of what they either did not read, or did not understand? '[A]ttacking' simply being an incorrect interpretation of the motives involving such selective quoting, according to you.

That come-back was a little harsh, but yes, I think Tyler actually reiterated your message.

Rather than the Mexican model of pensioner perfection, I prefer the Norwegian model, where reduced income inequality has made it possible for hands-on care to be a pretty well paying job, as I learned from reading the informative article about guest Swedish workers in Norway.

The economic answer to the robot problem lies in the book "I, Robot" by Isaac Asimov.

It seems to me you wind up with a First Law paradox. If a robot doing his job will put someone out of work, they are harming that person. But a robot not doing his job not only contradicts the Second Law, and the loss of the robot's productivity harms his owner, violating the first law. And I suppose, a robot that refuses to work will eventually be shut down, violating the Third Law.

Which leads me to the thought, in positronic programming, are corporations people?

"Harm" is an interesting concept. Would a robot have a different definition of harm for a retiree, than say a relative or a doctor? Perhaps a robots idea of palliative care might be better for the patient themselves, but not appear harmless to an ICU nurse.Would a DNR notice push a robot to defend a dying patient from a relative?

I think the First Law would have to be limited to direct, positive harm. If a robot were to try to avoid any negative second or third degree consequences of its actions, it could never do anything at all.

To your example, what if a robot doing his job puts someone out of work, but the same robot not doing his job results in lower returns on his company's stock, which means a lower return on a state pension, which means some pensioner goes hungry. Humans can't really deal with contrafactuals, let alone robots.

From Terry Pratchett's "Going Postal" (Mr. Pump is a golem, which is about as close as fantasy stories get to robots)

Moist Von Lipwig: I'm just a con man!
Mr. Pump: You have killed 22.8 people.
Moist Von Lipwig: I've never so much as drawn a sword.
Mr. Pump: You have stolen, embezzled, and swindled. You have ruined businesses and destroyed lives. When banks fail, it's not bankers who starve. In a thousand small ways, you have hastened the deaths of many. You did not know them. You did not see them bleed. But you snatched bread from their mouths. There will be no running.

There you go, with the IP "owner" concept.

What's up with the repeated IP owner references. Does Cowen have a reason for thinking ownership of land and other natural resources won't be a factor?

The sentence (A) "robots concentrate wealth in the hands of IP owners [in the future]" sounds just like (B) "industrial machinery concentrates wealth in the hands of industry owners [in modern days]". Doesn't it?

Simply put having less future workers seems like a soultion for the long term future decrease demand for workers. Notice all ten of the CNBC ten most competitive countries (Singapore, US, Germany, etc.) now have below replacement fertility so it the poorest least developed and disfunctional nations growing the population. Yes, this creates major long term demographic deft deflation spiral and the potential increase concentration of wealth with the potential of a major war.

Who says creative destruction was suppose to go smoothly? You write about the 1930's having lots of creative destruction.


Japan is way ahead of us in terms of an aging population. They also have lots of automation. Are there worker shortages? No there is a whole generation of underemployed people. And deflation. When people age they do not just drop out as workers, they also drop out as consumers.

Read the book "The Lights in the Tunnel: Automation, Accelerating Technology and the Economy of the Future"

It is all about the impact of robots/automation on jobs and the economy. It makes the basic point that machines do not drive consumer spending. As more jobs are eliminated, could a smaller and smaller percentage of the population sustain consumer spending and the economy? Would we want that?

Whether you agree with the book's conclusions or not, it raises many important and fascinating questions. Questions that are are really not being addressed elsewhere. Check out the reviews on Amazon (over 100 with many strongly opposed views).

If the robots can lower the costs of consumption, then they really could allow a smaller actual consumers to keep up consumption levels.

Underemployed is vague term that needs to be defined concretely to have meaning.

One of the problems with servants for an aging population is that US citizens are resistent to outside help. My mother was infirm the last ten years of her life, and we wanted to get someone to come to her house a few days a week to help her bathe and maybe clean a little, mostly to take pressure off my father. They basically refused, as they didn't want a stranger in the house. I was speaking to someone the other day who is going through the same thing with her grand parents. So even if you lower land costs so servants can be cheaper, people still won't use them.

True. However, I don't think this insistence on independence is immutable or a given; in fact I suspect it's a modern affectation. For most of human history wealthy people have not only accepted household help, they've demanded it. Something which struck me when I watched Downton Abbey - he has a servant tie his tie for him!

Also possible that this attitude is peculiar to Americans, and Japanese (and others) wouldn't be so resistant.

" They basically refused, as they didn’t want a stranger in the house."

I doubt such objections would apply to machines. So robots would be a win-win in this role.


if you want to take an argument to ridicule, why don't you write the argument yourself rather than deliberately taking something I wrote out of context? If your point was to say that it is ironic that we can both have rapid productivity growth and still have much of the population burdened by the taxes needed to support a growing population of elderly why didn't you just say this, rather than pull a section out of a statement by me, where the fuller context shows that I both understood and explained the irony.

Since your approach took much more time than the route I suggested, and served no obvious purpose, I did assume that the point was an attack on me. I'll plead guilty to assuming that there is a non-zero value on your time and therefore there was a reason you chose to waste it in this manner.

Can I repost an earlier comment?

Brian Donohue November 30, 2012 at 9:50 am

I’ll ask again: in what way is the “what are we gonna do with all these people?” worry not fundamentally incompatible with the “where are we gonna find the people?” worry.

I don’t think it’s fair that people get to worry about both of these things, but I do my part to make up for it by not worrying too much about either.

Tyler, the distribution and taxation 'problems' you raise, are these not issues with which society has had to grapple from time immemorial? I've fallen for 'but this time it's different' too many times in this lifetime. I ain't sweating this one.

But these both would not be the problem. It would be a dysfunctional governance system. Sure, it looks like we are already there, but I have more faith in the democratic process, not over the short term, but the long, in the ability to solve these problems.

This entire line of reasoning assumes too much homogenity of people and tasks. The same sort of thinking would mean one could say "it is not possible to have any meaningful level of unemployment and also have a labor shortage." But such shortages have been the norm in elite parts of the software world for decades. Why? Because unemployed house builders aren't instantly transmorgrified into primay care physicians just by going to the unemployment hall in some rural county with a Dr. shortage.

I predict that *both* of the things Mr. Baker says cannot happen together *will* happen together. In fact, the increasing inability of anybody non-technical to have a job with net positive tax payments will aggravate the financial pressures on the retirement system.

Partly this is because the interesting ratio isn't "old people to young people" it's "retirees+ children+the infirm to producers of goods and services". Some large number of "unemployed" people of course produce lots of good and services (raising children and taking care of infirm relations for example), but that doesn't solve the tax problem.

And even ill elderly people require a steady supply of non-healthcare economy resources (food, water, heat, housing, etc.)

Taxation will border on impossible - no matter what tax law says, confiscating the water pump "robot" or the sewer treatment "robots" won't actually solve your problem. The same becomes true with energy, agriculture, etc.

"OK, we'll tax the robot makers!" - right, that's like saying we'll solve a food shortage by taxing farmers.

Historically most members of society made meaningful contributions to the economy and so distribution and even redistribution could work. If we reach a point where 10% of the population has technical control over most of the goods and services the other 90% use, it will be the 10% not the 90% who make the rules.

[To see this, stop thinking about "ownership" and think about "effective technical control". So Jo Wise owns and programs the farm robot. You are unhappy with Jo and seize the robot to redistribute its output. You then find that the robot won't actually function for more than a few weeks without Jo's ongoing positive, active, support. Imprisoning Jo or doing anything at all that makes him less happy or less effective means less food. In the end, it is Jo who wins. And his "oppressors" who learn to live in mortal fear of Jo's dying.

If you have never managed a large software project, you probably don't grok how bad this problem can be.

I suspect that going forward everybody in society will come to understand it rather too well...]

Maybe the robots are the result of taxation on human laborers. If a bit less that half of the proceeds of your endeavor get strained through a government, why bother?

Has anyone done a study comparing the actual control and intrusion into day to day economic activities and decisions between the failed centralyl controlled empires of the 20th century and what we have now? Maybe the only difference is the freedom available to circumvent the control.

Isn't the AI going to be smarter than the humans? Telling people to go into STEM fields will be like telling John Henry to hit the gym.

I disagree on the politics in that situation. We are already in a situation where a majority of the population is technically not in a labor force, but they aren't weaker politically compared to workers. Many, such as the elderly today, have extra political clout because of more time to vote and organize.

@Brett - what you observe is in fact clearly true today - but today politics still works (and perhaps it always will.)

Reality probably will not completely catch up with my imagination, but consider the following end case, and how close to that end case the real world might come. (End cases all motiviated by "a very few % of the population with effective control of robots that produce most essential products end up with great power".)

End Case: One person, Jo, is the only person on Earth who can operate the FoodRobot. And due to climate disaster and other factors, all of the food on Earth comes from the FoodRobot. I claim that at that point, the 5 billion to 1 voting differential doesn't matter - Jo still wins all arguments.

Now, you might (and should) quickly point out how extreme that is, and implausible to last for any great time since Jo will die. On the other hand, I've personally seen (been fired from) small companies that couldn't go on without Andy or Sara, and when Andy died or Sara left, the company collapsed. The 7 to 1 or 23 to 1 votes didn't matter.

Consider another example - random reading on the web informs that in the 19th century, what is now Paraguay engaged in war which left them with a much smaller population and a very small percentage of able bodied adult males. As in able bodied males where 10% of the adult population rather than the roughly 50% that would be normal. In such a circumstance one might imagine that the much more numerous adult females would control politics. That doesn't appear to be what actually happened.

The Short Point - the political balance of power depends on some set of real world base rules. If the world changes so much that those base rules are profoundly altered, politics will be too. And in fact, democracy depends on a relative flatness of physical, technical, etc. powers. Robotics and AI and an ever smaller part of the population providing an ever larger precentage of essential products may alter those rules.

There is a lot happening in robots to aid the elderly. Japan is in the lead as mentioned above, but this is happening in many other countries as well. It's hard to tell how these changes will evolve because most are still experiments. But already you find robots designed to play the role of pets and helpers in ways that fit the emotional needs as well as the physical needs of their users. It starts with simple things like the smart pill containers that keep track of prescriptions to compensate for forgetfulness about whether you've already taken a pill or when you will need to take the next one. The video-phone/tablet combination has become an important part of maintaining social contacts and checking out remote living situations.

Some of the better designed more sophisticated robots are already treated as pets. Experimenters have learned that when a robot needs repair the users will not always accept another as replacement. There is sometimes a demand that this one be fixed and returned. People have grown attached to "their" robot. This is in large part because the design is accommodating the appearance and behavioral needs of the users. Behaving like a loyal helpful pet creates a comfortable relationship between robot and owner.

When the US was a capital exporter, our tax laws favored borrowing. Tax laws also favored investment in and financing of capital. Necessarily, by that choice, we taxed labor more relative to capital.

Yet, investment in human capital is just as important as investment in plant and machinery. Do we favor investment in human capital? Do we tax human effort more than we do capital (plant, equipment, derrivatives, etc.).

I think it is a fair question to ask: should we lighten up on taxing human labor and employment, and tax capital more. It all goes into the same pot. And, don't think we don't use taxes to benefit capital, either: we have a system of roads, waterways, a military protecting sea lanes, etc., and the GE's of the world don't want to pay for it, although they benefit from it, and are often suppliers to the government.

Two points:
- labor's share of GDP is much larger than capital's;
- capital is more mobile than labor (I'm talking about international mobility, medium to long-term).

The share issue is irrelevant, and we are talking about marginal changes. Second, given that machinery is mobile, and people, for the most part in the US as between the US and foreign countries, are not mobile, this would be an argument for investing in human capital here. I am sure some foreign government will subsidize capital investment, but human resources could be qualitatively different, making a move less likely. Moreover 75 percent of trade in goods and services, or more, are produced here and consumed here.

Growth can come from improved human resources, just as it can come from capital.

Many american pensioners are moving to Mexico

I am not concerned that excessive productivity growth will leave tens of millions without jobs

My problem with the quoted section (and the rest of the original post which I read in full) is its 'gotcha logic.' It reminds me of the 'if you didn't think stimulus would work, then you can't be worried about the cliff.' It's not an endearing type of rhetoric (even when you agree), especially in macro where there a free variables flying all around. Tell me why your argument is right, not why the other guy's is stupid. I agree (as it seems everyone else does) that distributional issues are important to consider, but even in real-time macro that turns into a huge mess...and now we're going to project forward with more robots and other stuff. But what ended up bugging me about this post was the utter fail (to no one's fault) of this format. I thought blogs with their flexible, arms' length, open format would 'foster a culture of open exchange.' But this was not an exchange, it was like an intellectual snow ball fight. Kind of fun but kind of sad too. Seminars can be like that too, but there's usually more closure by the end. Oh well, technological innovation cannot solve all our problems.

"I agree (as it seems everyone else does) that distributional issues are important to consider, but even in real-time macro that turns into a huge mess..."

Sounds kinda Hayekian.

Certainly a repudiation of Lenin's 'piece of cake' notion: "All that is required is that they should work equally, do their proper share of work, and get equal pay; the accounting and control necessary for this have been simplified by capitalism to the utmost and reduced to the extraordinarily simple operations--which any literate person can perform--of supervising and recording, knowledge of the four rules of arithmetic, and issuing appropriate receipts."

"Sounds kinda Hayekian." I sure don't hear that often in response to my arguments, especially ironic since the point I was trying to make was formed in a work-horse Keynesian-ish environment. I was not talking about redistribution, I was talking about how shocks to the economy are distributed across households/firms in the economy. Macro economists tend to look at aggregate trends in shocks...I am sticking with the idea the productivity growth is in general a good thing. And yet, of course, there are times when productivity shock could be bad for the overall economy if it concentrates in a new and unfortunate way. Of course economists need to think through all these possibilities (the huge mess comment comes from trying to do that carefully with some recent shocks) and not just go with the historical, aggregate average. Wasn't trying to be obtuse with that sentence (though I may have made it worse now)...

Hi, I just dropped by to view this website. It looks really full of great
content and I liked browsing it, thank you for the helpful

This is fantastic - 'In reality, many old people would fare somewhat better if our economy were somewhat more like that of Mexico, namely with cheaper land and cheaper servants.'

Maybe instead of servants, we could just use unpaid unemployed people, so they can gain skills.

But honestly, this is just a fantastic satire site, because really, no one else would believe that the general director of a center dedicated to the free market would be so straightforward in his hopes for what that market could provide for older people - more servants.

Especially in light of the fact that he is still apparently childless, meaning he may need someone else's children to provide for him in his later years. Hopefully, it appears, as servants.

This is seriously one of the most laughable blogposts I have ever seen. Trying to catch Dean in a contradiction that he has written about thousands of times -- and, in fact, that he has written an entire book on ( -- is like calling out Ken Jennings for misunderstanding history. I understand that Cowen probably doesn't read Dean's work that much, but he should if these are the types of cheap shots he's trying to attract readers with.

The idea that technology decimates opportunities for employment has always struck me as pretty fanciful. Yes, the cotton harvester put a lot of cotton-pickers out of work, but we seem to have survived all right. Is anybody calling for Solow to give back his Nobel? Is the premise that owners of capital will just sit on productivity gains and fill swimming pools with gold coins like Scrooge McDuck? Instead of the Luddite fallacy, I'm going to start calling this the Scrooge McDuck fallacy.

Internal combustion engines doubtless put a lot of blacksmiths out of work, until they learned how to do tune-ups and oil changes and balance and rotate tires. Robots still have to be designed, programmed, constructed and maintained. You still have to mine and refine the materials and get them to the robot-building place. You still have to get kilowatts to the robot-building place. You have to build the robot-building place and keep it clean, or is the future just robots all the way down?!!! The mind boggles. Train your daughters in archery, people, because The Hunger Games are your future.

The real problem is the externalities we impose on the process of labor re-tooling.

There is also the problem that you can only tax people so much before some start working in the non taxed sector (mostly wives drop out of the taxed economy first and produce for in family consumption). This makes it difficult to fund the expansion of welfare programs to cover those displaced by machines despite the rising productivity that the machines create.

Of course I do not agree that machines will displace workers fast enough to create a big problem.

Comments for this post are closed