« On "letting the market set interest rates" | Main | Migration and Population Size »

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Brilliantly said Nick! Computers make my life better 95% of the time, and in the other 5% they reduce it to a cesspool of unmitigated misery. My own pet peeve is websites that require me to have passwords of a certain length, or with a certain combination of capital letters etc., which guarantee that I will never, in a million years, remember what the password is.

Nick, great post; Woolleylaw, great comment.

This kind of thing is really important to think about when it comes to things like measuring poverty or social exclusion. It's hard to find a payphone these days except in older neighbourhoods/buildings. Some apartments aren't even wired for land lines. So basically if you don't have a cell phone, you're stuck. It's all very well to say "you can't be poor! you have a smart phone for goodness sake!" But it's got to the point when a person who doesn't have smart phone (or a cell phone + a computer) is excluded from society.

b.t.w. I do check the spam filter, but most of the time it contains nothing but spam, so it's a bit tedious to keep on doing so. I'd really much rather you just sent me an email...

Well, one thing they will do is get health insurance, if they live in Canada. Or in any other rich country that isn't the US.

It's possible this is a problem that can be solved.

Well Nick, if you really want to feel inadequate you should try and set up a Smart TV.

They will be fine, always have been. That people are unwilling to adapt is not solely province of the meek.

Google+ confuses even the young and educated. It's noisy and akin to social networking on ADD. Don't worry you don't "get it".

You can sign up for Obamacare by phone.

JW: "Well, one thing they will do is get health insurance, if they live in Canada."

Don't be so sure. Listen to this:

I have really good dental insurance through Carleton. Supposedly. I went to a different dentist, and they didn't have me on file (no "cookies" on their computer). I told them the ID number for insurance, but it had the wrong number of digits, so we couldn't get it to work. So I had to pay cash. Now, supposedly, I "can" download the forms from the web, if I can find them, and submit a claim for reimbursement. But I just can't face trying to do it. The dentists's receipt is still sitting on my kitchen table somewhere, I think. But it's probably too late now.

Do I *really* have dental insurance? No, I don't. Only people who can and will work the system have dental insurance. (And insurance with a $1,000(?) per year cap is really really stupid anyway, because people need insurance against catastrophes, not against small stuff, when the odds are actuarially unfair by necessity of moral hazard and transactions costs). Just like Carleton's "professional expense reimbursement" forms, which I haven't submitted in years. Just like Quebec's $7 (whatever) per day daycare, *if* you can work the system and get a place for your kid.

Yes, Canada has good healthcare insurance, *if* you can work the system and find a family doctor, and can sit there waiting. And what happens when they require you to "login" to get treatment, and you can't remember your password?

Some of us are really useless at making the system work. And lots of people are worse at it than me, and it matters more for them than me.

JW - Nick is right on this one. Even for "free" health care one has to have a health card, and get it renewed every few years. In the past I've gone to the wrong place to get my health card renewed, or ended up not having the right forms, or waiting for hours because I couldn't work out the complex system for calling out numbers.

Nick, the deadline for dental claims is 18 months - if it's more than 18 months, chuck it, if it's less, bring it into work next week. On the PER stuff - you know that MEC will do it for you if you look helpless.

Dental insurance is particularly evil for students. I don't know why they don't rise up rebel.

Via twitter:

@franceswoolley @stephenfgordon Would have commented, but couldn't remember password on Typepad, OpenID, and won't use FB or Twitter for it.

— Tom Slee (@whimsley) November 2, 2013


Getting ahead in all walks of life is all about working the system. The private sector is the same, if not worse. If you can't work people, you're dead in the water. People are just another type of 'system'. And having no access to the people who can do the stuff you need done, or being crap at working people to get what you want doesn't seem to illicit as much sympathy as not having a cell phone or being crap with computers.

And I totally agree on the dental thing. I tried explaining that to co-workers once (pay for your damn cleanings, what you want is insurance that covers implants). They didn't get it.

They won't get it. Compare Disability Insurance to Dental. Dental receives a sizable tax break, the premiums are tax-deductible and the benefits are received tax-free. It's pay by another name, using the government's dollars.

LTD does not receive any tax break. Either the premiums or benefits are taxable (both forms exist) but not both for the same policy. Dental covers minor expenses, DI covers catastrophes.

You're not alone. Here's a computer professional with similar complaints:

http://arstechnica.com/information-technology/2013/11/its-the-little-things-how-small-conundrums-make-many-hate-computers/

st to check if Typepad will reject me once again: Nick: it's not you or me. It's horrendous programming ordered by managers who don't give...In a second screen in the background, I play reruns of showa I missed on Radio-Canada and Télé-Québecshows on to.tv and telequebec.tv. No problem, I stopped even trying to do the same on cbc/natureofthings. I wrote and didn't even receive an automated reply

so it just doesn't happen to me it happens to brainy people as well from Canada no less.

I fell your pain

The big problem is not with having to use your computer to collect Social Security or buy things. As technology and artificial intelligence advances it becomes easier -- compare computers today to MS DOS, or even earlier. The really big problem is as computers and robots get smarter and smarter, they become able to do more and more jobs for pennies per hour compared to a human. So for many people, yes, the free market will give you a job, for 3 cents per hour, because that's what makes you cost effective compared to a computer or robot doing it. What's really worrying is what about the people with learning disabilities, or even just those with only average intelligence? The Republicans don't even want to feed them right now (slash food stamps!), let alone give them a roof over their head and health care. Are they supposed to die if they can't find a job that pays enough to stay alive? This is already a serious issue, but wait another 10, 20, 30 years. For a great article on this issue by Kevin Drum, see:

http://www.motherjones.com/media/2013/05/robots-artificial-intelligence-jobs-automation

1. With respect to typepad, it is also for me a frequent experience, that typepad does not work. Some blogs dont like my "genauer", here it often just does not post things,
even when it displays: your comment was posted.

It seems to help to immediately do a page reload after posting a comment, but not always.

2. For health care in Germany you have your "Gesundheitskarte" "health card", with name, photo, microchip, insurance-ID. You show it to the Doctor and that is it. No cash flows, no disimbursement form, no caps, nothing.

I dont see an alternative to that some folks think up how to work the system efficiently, convenient, and safe enough against too much abuse. But, obviously, it can be done.

Richard. It's not obvious to me that computers will cause real wages to fall. It might happen, or might not. My old post on the subject.

From Determinant's point about dental insurance being a tax-free way to pay salaries: that probably explains why we have these very inefficient types of insurance. Which means it's a bad tax policy. Probably worthy of a post (but maybe not by me), making the point that insurance must be actuarily unfair, and that it's therefore efficient to have insurance only against big losses. Not an easy point to explain in words.

(But the tax-free pay thing can't explain why students get dental insurance, can it??)

Thanks for all the other good comments, and kind words. I don't have anything useful to add.

you are now my favorite person on the internet, even if you don't know how to use it.

And of course, being dependent on the Internet also means you must pay the cost of access to the net, a cost that can reach $1000-2000 a year, not including the price of the equipment. For homeless people using iPhones because this is the cheapest way to access email, telephone and general Internet entertainment, it provides huge benefits, but I wonder if that cost per year might be for many people the reason they cannot save for rent and a damage deposit?

But to get back to your original point, I think of the situation of my elderly mother and the son who lived with her till her passing last year. I made sure they had net access in the house, but I needn't have bothered. My brother is disabled with depression and learning problems, and never did become even slightly comfortable with the system. Our mother was blind from cataracts. So how on earth would the two of them get signed up? I imagine the already overworked county social services departments would be the ones expected to accomplish this. You remember them, the people whose funding has been slashed so deeply in the past 15 years?

No idea how they are managing, but geez Louise it would save billions to just give everyone a health card and be done with it. Sigh.

This is amusing, but doesn't seem like a real problem to worry about.

Human society has coped with such situations many many times. Similar worries would have accompanied the adoption of fire, animal herding, agriculture, nation states, writing, money, equine based transportation systems, plumbing, electricity, telephones, automobiles, typewriters, microwaves, CDs, DVDs, and even computers.

People do fine without the Internet if they don't have the time or inclination to learn what is needed to use it beneficially. Some people don't have cell phones, some don't drive. There exist alternative ways to shop, communicate with friends, and to sign up for government programs. None of these require the Internet.

If you email the blog owner sometimes they run a thread mentioning your comment. You could email the posters you want to participate, and arrange to post the email running conversation on a blog somewhere. Someone would certainly make a thread out of it. My problem is, certain members of the cdn blogosphere covet my future wife and moderate me.

I've done posts complaining about the insane "dental insurance" in America, and I got sucked into Google+, don't know what it is, and never respond to anyone. And I often have trouble leaving comments (not here fortunately.)

So I strongly identify with everything you say.

I went to Italy, logged on a bit, and WordPress decided I was a spammer. Fortunately, a simple use of a different email address solved that. But yes, petty and not so petty annoyances seem to be built in, and more so rather than less. "Upgrades" that really, really aren't are a bane of recent I-tech existence.

Not that I don't like I-tech. I do, a lot. My ability to access scholarly works easily and cheaply has been hugely expanded.

Nick wonders:

But the tax-free pay thing can't explain why students get dental insurance, can it??

Parents become conditioned to paying for dental through insurance, and don't realize what it costs. So they look for "dental insurance" for students. Add in insurance agents to have every incentive in the world to sell policies, and you have the conditions for this sort of sale. Plus you have the small group of students to who don't have dental coverage through their parents, and so they feel left out, rational expectations be da**ed.

It's part of what I call the "Expected Social Contact"; the hazy picture of health, dental and pension coverage that a "decent" employer is supposed to provide. Group insurance agents spend their lives trying to make this into reality, or some vague imitation of reality, some of what they sell is positively dreadful.

Nick, thanks for the reference to your robots post. This is a subject I am very interested in.

I think it's a good initial effort, but it's missing seriously important things.

Let's first consider your part I, "Labor and Robots Only". I've basically thought about this scenario. The main idea is that humans and robots can just do their production thing independently. Robots can produce a bazillion goods cheaply without humans, but so what, the humans who don't own robots will still set up their own 1950's factories and produce the 1950s amount of consumption goods, and have the 1950's lifestyle, in fact better, because they can trade their consumption with the robot owners for increased utility. These humans still have the ability to work on the assembly line independently of what the robots do.

The problem? The people who own the raw materials (your N) aren't going to let it be used by the 1950's workers to set up a 1950's factory. They're going to take the same wages that they would have paid 1950's workers, and use only a tiny fraction of it to buy robots to turn all of their raw materials into goods. So, the 1950's workers won't be able to keep doing their thing and keep collecting their same wages, because they won't have the raw materials, the factory, the skilled managers and engineers, and other inputs, who will now go to the robots' facility.

Your part II: Basically it comes down to this: If robots keep getting to produce more and more (of the single consumption good in your model), then you get a bigger and bigger return from forgoing your consumption and buying a robot. Today 100 units of the consumption good may be used to buy (or make) one robot, which can make me 5 units of the consumption good per year (say forever, WLOG). In 10 years, with improved technology, the 100 units of consumption to make a robot may make me 10 units per year. For there not to be an arbitrage, or an abnormal risk-adjusted return, the real market interest rate would have to go up from 5% to 10% (assuming your g is insignificant, which you pretty much do over the long run).

So, the question is, why aren't interest rates rising? Well, your model is very simple, with just one kind of interest rate, or return. I'd look at stock returns, where companies buy the robots that produce more. And also, I'd say, the arguments of Kevin Drum, Noah (He should really go to one name. It would be great for his career.), and Brad Delong, are that this will take off in the future exponentially, not that the huge take off is here now (although the "neck" of the exponential take off is starting – see especially the Drum article I cited on this.).

At that time, we may see stock returns take off, and interest rates follow, if risk aversion doesn't go up a lot at the same time. But even today, in recent years, earnings have been very good, and stock returns have been awesome, since spring of 2009.

Also, your model has only one consumption good (or an aggregate consumption good). Robots can plummet the costs of producing some kinds of goods, but not others. Suppose you add to your model two kinds of goods, a and b. Robots take off and produce vastly more of type a, per robot. The return on robots in type a goods skyrockets, but in dollars, maybe not, as the price of type a goods plummets perhaps as fast as the production per robot takes off. The price of many manufactured goods, in dollars, and relative to other goods, has, in fact, plummeted.

Here, I think position/context/rank/prestige may matter a lot. Think of – using Robert Frank's terminology – positional and non-positional goods. Robots, ever improving, can produce exponentially more and more non-positional goods per robot. But positional goods, by their nature cannot be produced more ever per capita. There can only ever be 1% of the top 1% of homes or cars. If you want a home in a top 25% neighborhood, and to drive top 25% cars, how much robots produce will be irrelevant. The more and better homes and cars they produce, the more and better home and car you'll need to achieve your need to be top 25%. If robots double productivity at making a house, you may need twice as big a house to be in the top 25%, or more expensive and rare raw materials in it, granite, mohagony, platinum,...

Robots may be able to produce horsepower and leather for a car at half the price of 30 years ago, but now you need leather and horsepower that much more to achieve your desired level of prestige, or perceived "quality". Positional externalites – so profoundly powerful in the real world – shatter the simple consumption good and utility function common in economic models, and turn things upside down.

The final scenario you have, "Labour and Robots plus Land.", does show the nightmare scenario! But you kind of tut-tut it, like, ok, if you *have* to throw in land (or raw materials, or other needed inputs of production), *then* you're right. Well? Don't we need these things? That is how it is, so there is, potentially, a very serious problem. Yes, if laborers could produce the consumption good (or the consumption goods they need) with just hours of work alone, without the need for raw materials, skilled workers, land, and other inputs, then yes, they'd just keep doing their thing and producing and consuming just as much, no matter how productive the robots got. But that's not much solace to people who live in the real world.

And that exactly relates to my 1950's thought experiment at the beginning of my commenting. The unskilled manual laborers on the assembly line can't just keep doing their 1950s thing and making just as much of the consumption goods, independent of how much the robots produce, because no one will give them the raw materials they need – it will go to the robot factories. No one will give them the highly skilled managers and engineers they need. And so on. All that's going to the higher and better use, at least, as far as production, in the robot facilities.

Consider a person who can only pick fruit. He can pick 1 bushel per hour and makes $15. Then a robot comes along that can pick one bushel for only $1. Now, the average income of society may go up, but this guy's wage is going from $15/hour to $1/hour. You may say, if he can pick friut, he can also do other, different manual labor. But what if, due to exponential advancement in AI and robots, this happens for *every* kind of manual labor, exhausting every kind of work he is currently capable of doing?

As you write, in this third and most realistic scenario, "With a little bit of math, we can show that human plus robot workers earn a constant share b of total output, and landlords earn the remaining constant share (1-b). But as more robots are built, and the robot/human ratio K/L rises, human workers earn a decreasing share of b."

In other words, GDP goes up, but so does inequality, and it skyrockets as robot productivity skyrockets, with workers earning diminishing less relative to owners of raw materials, or capital. It appears your model admits the problem.

As Drum writes:

"The machines that people like Cowen and I are predicting for the 2030s just flatly have no analog to previous machines.

Those machines won't need help from ordinary humans. In fact, as they get smarter and smarter, they won't need much help from really smart humans either. Eventually, they won't need any help at all. Past machines always did, and that's the decisive difference. If you wave this away, you're missing the whole debate. You're pretending to argue without actually addressing the main point of the techno-optimists: What happens to human labor when machines are smart enough that they need virtually no human guidance at all?"

At: http://www.motherjones.com/kevin-drum/2013/10/yes-technology-going-destroy-middle-class

In this case, income gets superconcentrated in the hands of the owners of capital. And Brad Delong makes the similar points here:

http://delong.typepad.com/sdj/2013/10/trying-yet-again-to-get-a-proper-frame-for-tomorrows-uncharted-robots-conversation.html

Also, later I'll pull the cites, after I get some sleep, but the percent of income going to capital, after being stable for a long time, *has* in recent decades gone up a lot -- in line with your model.

This is actually an interesting post. I have just two things to add:

1) Your problem: I searched a little bit about TypePad and I think your account or your IP address are probably now on a so called blacklist of their "social antispam" service. How did this happen? With antivirus/antispam protection it is mostly a technological secret. You may have been infected and malware could send out spam messages. Or you may had a time where something about your comments may have seemed as fishy and you got blocked.

The best course of action is for you to contact TypePad support and ask for being whitelisted. If you actually purchased their platform you should have better chance. But there is no guarantee! I know of a company who got into serious trouble for one ill-advised email campaign. It took them some time not to have all emails sent from their email infrastructure marked as spam.

So in short - there is a war being waged out there on the interwebs. Think of it as a Wild West. Sometimes innocent bystander gets caught in the firefight for no reason and there is no reasonable think he may have done to prevent it.

2) Broad topic of technology scare: I think you overrate it a little. I personally am a user of technology since I was 6 playing with ZX Spectrum. I built a few computers myself and I know a little bit of everything from coding to disinfecting malware; mostly self-tought. And believe me, computer technology was far more complicated back in the old days. I remember some people on college programming their own drivers just to be able to run some Linux distribution on their hardware. Compiling kernel was as usual as going for a beer with friends. It was a long way until things like Ubuntu. But most importantly we are still not in the final destination yet. IT is not really mature - especially with real internet being basically only few years old. Even with far simpler technology - like bicycles or cars - it took decades to have them properly used. I think that it may not be until mid-century when we will have most of the things sorted out with our current level of IT; and then maybe we will have different troubles like for instance how deal with advanced AI systems.

So don't feel bad. I know a few 20+ people who are really bad with technology - and they are still sucesfull in their respective fields. Just because somebody thinks that he knows how to use some Office software or he knows how to upload picture on the Facebook it does not mean that he knows anything about other fields. There will always be people like this (unless humanity is involved in some heavy gene engineering) and as long as this is a fact it will be IT industry that has to serve these users not vice versa.

Richard: good comment. You say:

"The final scenario you have, "Labour and Robots plus Land.", does show the nightmare scenario! But you kind of tut-tut it, like, ok, if you *have* to throw in land (or raw materials, or other needed inputs of production), *then* you're right."

Is that how my post reads? I didn't intend it to read like that. I think "land" is very important. I think you can't understand the effects of robots without having land in the model. I am tut-tuting the people who talk about robots, especially people who talk about nightmare scenarios from robots, without talking about land. (And the people I would really tut-tut are the people who talk about the *general* equilibrium effects of robots using *partial* equilibrium reasoning.)

If robots drive down wages, they must drive something else up. Otherwise we can use the argument of your first big paragraphs (the "1950's" thought-experiment, which I remember Mark Blaug using against the Marxian argument of falling rate of profit plus increasing immiserisation of the proletariat) to show it can't happen. Now, what is the thing they drive up? Interest rates, or land rents? Empirically, "land rents" seems to be the answer to that question, at least so far.

Kevin Drum's piece says *nothing* about this. Brad DeLong's piece says a little, but only by implication.

It really bugs me that people can write screeds of stuff about the effects of robots on wages without talking about their effect on interest rates and land rents. Hopeless hopeless hopeless partial equilibrium reasoning. Massive fallacies of composition sticking out like a sore thumb. "If cheap robots could replace professors then professors' wages would fall. Therefore if cheap robots could replace everyone, then everyone's wages would fall!" TOTAL BULLSHIT!!!

Let me stick my neck right out, and blow my own horn: I know next to nothing about robots. But because I *don't* commit that appalling fallacy of composition, and instead build 3 little GENERAL equilibrium models of robots, I think my quick and dirty stupid little blog post is way better than all the other crap which doesn't do that!

I don't get your bit about positional goods. Sure if all goods were pure positional, then an improvement in technology doesn't make people better off. But I don't get how that relates to robots in particular, and wages in particular. It would also be true if land were the only factor of production, and there were productivity improvements in land. Land rents would rise, but landlords would be no better off.

Hang on! Something's happened! I think maybe Typepad have whitelisted me, or something! Because I managed to post a comment on Mark Thoma's blog. And now just managed to post one here, under Nick Rowe with a space! (And I think I managed to re-set my Google password! Fingers crossed...

YEP! It's working!

A big THANKS to whoever did that for me.

BTW, the people I really ought to be blaming for all this are the spammers. Why can't some of those teenage hackers do something socially useful for once, like taking down all the spammers?

Assume robots have enormous productivity. Then, if the marginal revenue equals the marginal cost, and the cost is effectively the cost of the raw materials, energy and rents, the suppliers and the robotic producer have a closed cycle economy.

Meanwhile humans have their own economy exchanging human produced goods and services. The question is how these economies connect. Humans need to buy consumer goods made by robots. To do this they either need to perform human services for participants in the robot economy, or there have to be transfer payments.

Of course, if robots were inexpensive, most people could own some and do their own production of at least some consumer goods.

And so, we would have something like your model with land. It's hard to see how the rentiers could be paid, since they could easily get their own robots and do their own production. Humans could only get rental goods by performing human services for rentiers.

So we end up with a political problem as much as an economic one, unless technology finds a way to make the inputs inexpensive. This certainly might be possible for energy, and we already know some ways of substituting inexpensive inputs for expensive ones, like replacing platinum catalysts with ones based on iron. It's hard to go much further with this, since the unknowns get too large. A lot will depend not just on developments, but also on their order and rate of occurrence.

The economics of spam are rather interesting. This might make a good classroom example. You have an almost infinitesimal cost per email and an almost infinitesimally small response rate. Proposals have been made for solving the problem with a tiny charge for sending an email, say 1/10 of a cent.

Unfortunately most email is sent from botnets, so the charge would just let people know they've been pwned - not an entirely useless result.

Here's the cite I was talking about on the rising share of income going to capital:

http://krugman.blogs.nytimes.com/2012/12/08/rise-of-the-robots/

Quoting:

Robots mean that labor costs don’t matter much, so you might as well locate in advanced countries with large markets and good infrastructure (which may soon not include us, but that’s another issue). On the other hand, it’s not good news for workers!

This is an old concern in economics; it’s “capital-biased technological change”, which tends to shift the distribution of income away from workers to the owners of capital.

Twenty years ago, when I was writing about globalization and inequality, capital bias didn’t look like a big issue; the major changes in income distribution had been among workers (when you include hedge fund managers and CEOs among the workers), rather than between labor and capital. So the academic literature focused almost exclusively on “skill bias”, supposedly explaining the rising college premium.

But the college premium hasn't risen for a while. What has happened, on the other hand, is a notable shift in income away from labor...

I think our eyes have been averted from the capital/labor dimension of inequality, for several reasons. It didn’t seem crucial back in the 1990s, and not enough people (me included!) have looked up to notice that things have changed....

But I think we’d better start paying attention to those implications.

You're absolutely right, Nick, that you have to be careful with partial equilibrium here. Here is, I think, a nice, ultra-simple, general equilibrium model of robots:

There are two types of people: Farm owners, who also have human capital, and workers, who only own their human capital.

The farm owners need work to turn the resources of their farms into food income. They can supply their own work, or hire workers, so as to have more leisure. So we have a bargaining issue, where the utility surplus will be divided in some proportion between the workers and owners, depending on bargaining power and ability. But both parties will get more in return than what they give up, or they wouldn't make the trade.

Suppose that after the bargaining the workers agree to work 40 hours per week in return for 12,000 calories per day for them and their families.

Now suppose that the farm owners invent a robot that costs an insignificant amount to build, and can do the same amount of work as a worker, but only needs 100 calories per day.

Now, the amount of income the workers will be able to bargain for will not exceed 100 calories per 40 hours. Without food stamps, or a safety net, they and their families will just die. And we know what the Republican Party thinks of food stamps and a safety net. I think for many of their politicians the dying part is a feature, not a bug.

Total GDP will increase, especially if well measured to include the value of leisure, but obviously, income inequality will explode, and the wages of workers will plummet to almost nothing.

Next, consider the same model, but with two kinds of workers, skilled and unskilled. To produce food, both kinds of work are necessary. The skilled and unskilled workers both only own their human capital. The farm owners own their farms, and have human capital that allows them to do unskilled work, but not skilled work.

The skilled workers have much more bargaining power than the unskilled. First, the farm owners cannot substitute their own labor for that of the skilled, because they don't have the skills. Second, the work of the skilled creates much more food than the work of the unskilled per hour, and is essential so that the work of the unskilled results in production. After bargaining, the skilled earn 30,000 calories per 40 hours of work.

Now the farm owners invent the same robot as above. It can do the work of the unskilled workers at a cost of just 100 calories per 40 hours. So, the income of the unskilled drops below 100 calories per 40 hours.

But the robot cannot do skilled work. And, to add more production before required 40 hours of unskilled work to 1 hour of skilled, and produced 15,000 calories. That is to produce 15,000 calories required 12,000 going to the unskilled labor, plus 1 hour from the skilled worker, for a net of 3,000 from the skilled worker's hour. But now that there's an unskilled robot, the skilled worker's hour is combined with the robot working the 40 hours, but at a cost of only 100 calories. So, the net gain from an unskilled hour is now 14,900, as opposed to the 3,000 before. This gives the skilled workers a lot more bargaining power in wage negotiations, and so their income rises as a result of the rise of the robots.

Except...

One day the owners of the farms also invent a robot that can do the work of the skilled workers too, and for just 100 calories per 40 hours. So now the skilled workers suffer the same fate as the unskilled, with their income dropping below 100 calories per 40 hours. GDP again increases greatly, and income inequality explodes even further, with now almost 100% of GDP going to the owners of the farms, the owners of capital, and almost zero going to the people who were born with only their human capital.

Richard: Ah! Farmers! My sort of model! My family are all farmers, except me. And one day I must write a post about economies of scale and Coase and UK farming. Because it has changed so radically since my father's day, and nowadays my nephew is the last person for miles around who would still be recognised as a farmer in the old sense, and he employs nobody. Yes, your model sort of works for farming. There are people who own land, and people who own machines, and some people who own both. And they mostly work with their land and machines. But the old farmworkers who owned neither have almost disappeared off the land

I remember though that Paul Krugman did a follow-up post, where he tried to reconcile capital-biased technical change with falling real rates of interest. His solution was to say that some income that was measured as returns to capital wasn't really returns to capital. It was returns to those who invented new products and earned quasi-rents on those inventions. That was the "third factor", rather than land. He may be right, but I would still push for land.

Thanks Nick, this has helped a lot for me to better understand this issue. It's something I'm very interested in. My main career is in personal finance. I've taught 4,000 students personal finance at the University of Arizona and a half a million through my company National Personal Finance Education (and subsidiaries), a provider of a personal finance course that meets the US requirement for people in bankruptcy.

I'm very broad, flexible, practical, and pragmatic in personal finance, whatever related to money will better people's lives, substantially, and the most, including psychology, sociology, relationships, health, safety, and nutrition, and careers to name a few. And I have my U of A students do a research career plan, for their first and second choice careers.

So a big issue, a big thing I should know about, is how much of a danger will smart computers/robots be when these 20 year old U of A students are 30, 40, and 50, with families. And what are the best careers, and career plans/strategies, to minimize these dangers. And what are the worst (airline pilot doesn't look good with computer driven cars already).

So, thoughts on this current learning experience:

1) One of the good and bad and exasperating things about academic economics is that it's so hard and time consuming to do original work – and often to get published and do a dissertation – in that you rarely can just sit down and think about it and create something. You have to first, or also, see what's been created in the area, or sub-area, or sub-sub-area. And if you want to really do this well, this can mean massive time studying, and understanding, and fully covering, the research to date. We're sitting here pondering simple models, which has helped a lot to understand, and check internal consistency, but, of course, there may be many really good models already out there related to this.

Specifically, as Krugman said in the cite above, "This is an old concern in economics; it’s “capital-biased technological change”, which tends to shift the distribution of income away from workers to the owners of capital." So, I have to look at the capital-biased technological change literature and models – over the coming months and years, of course, as highly limited time permits. I take the long view on much learning. I'm forced to.

2) In doing my "farmers" model, a big thing that stood out to me: The unskilled workers could do nothing that robots couldn't. But really even the least educated workers are then "skilled" in this sense. There's still a great deal of productive things that they can do that robots can't, and there will be for a long time. The problem is supply and demand. If the number of unskilled (low education and training) stays as high as it is globally, and robots take more and more of what they can do, then there's less and less things left for this high number of people to pile in and compete to do, forcing down the marginal product of labor in that diminishing number of things ever lower.

Now, note that the skilled workers made more, a lot more, when the robots began doing the unskilled labor at a fraction of the cost. Essentially, you got skilled workers working *with* robots, greatly augmenting the productivity, the marginal product, of the skilled. Often the advice you hear from people who think about this is find a career or skills that it's hardest for smart computers/robots to do. And usually they say something involving high level abstract thinking, as opposed to mechanical/rote/repetitive things.

True, to a large extent. But another way to think about it, in addition, that's perhaps stronger and more valuable, is that you want to find a way to work powerfully *with* computers/robots as a team, like the skilled workers in my farmers model worked as a team with the first generation robots. You don't want to bypass robots; instead you want to find a career where you can be a valuable member on the computer/robot and human team (and of course a member whose role can't be duplicated just as well by a computer/robot).

I think of what medical doctors will do in 20+ years, with IBM's Watson and more, and I think perhaps, one MD will top manage a team of many smart computers/robots, nurse practitioners, nurses, technicians, and business people.

For careers what does this mean. For one, skilled computer/robot support and production of computers and robots jobs could remain high paying (when robots can 100% make robots for even the most powerful and best robots, then we're nearing the end of the need for human material production anyway).

3) Have to study the wage bargaining literature and what determines wages. Interesting thought – You often hear that water is incredibly valuable, because you would die in days without it, but it's so cheap to purchase a gallon from your tap because it's so unrare. But when you say water is incredibly valuable, you have to distinguish which gallons? The first ones are; the ten thousandth one per month is worth pennies or less to you. It's all about *marginal* benefit or product. The same with workers and wages. If there's too many people who can do what you do, and unless the returns are quite linear or little concave (and ask if they are!), you start driving the marginal productivity, and the wage very low.

Richard: Sometimes I have this deeply evil thought that perhaps civilisation only advances when ancient libraries get torched from time to time. Because it gets to the point where it's easier to reinvent the wheel than spending years in the library reading all the previous research on wheels. And that's one of the reasons I like blogging.

Ignore the vast literature on wage-bargaining. I think it's 99% irrelevant for this question. Work out what happens to demand and supply of labour, and the competitive equilibrium wage. Later you can ask how it changes if the labour market isn't competitive.

The literature on capital-biased technical change, and capital theory more generally, is a bit of a jungle full of warring tribes. Write down a rough model that you think makes sense to you, then pass it around, see what feedback you get, and see what people suggest you read, and see if any of those tribes will adopt you. (I would recommend reading Irving Fisher for a sensible and understandable big picture on capital and interest.)

I think the only thing you really need to do is make sure you have a *general* equilibrium model.

btw On Nov 4, I added "Nick Rowe" to Google+ and joined the "I am NOT a spammer, G-Da**it!" group. Your problems should now be resolved.

Anon: I'm not sure I understand all that. But if that works, I thank you very much!

The comments to this entry are closed.

Search this site

  • Google

    WWW
    worthwhile.typepad.com
Blog powered by Typepad