THE JOYFUL REDUCTION OF UNCERTAINTY

Technological Unemployment: Much More Than You Wanted To Know

[I am not an economist or an expert on this topic. This is my attempt to figure out what economists and experts think so I can understand the issue, and I’m writing it down to speed your going through the same process. If you have more direct access to economists and experts, feel free to ignore this]

[EDIT: Commenters pointed out that I was confusing the trend in total manufacturing jobs with the trend in percent jobs in manufacturing. This would make me less confident that there is no technological unemployment, except that I think the trend of percent jobs in manufacturing represents women entering the workforce and overwhelmingly going into nonmanufacturing jobs. Unclear that this requires a major change in conclusion.]

Technological unemployment is a hard topic because there are such good arguments on both sides.

The argument against: we’ve had increasing technology for centuries now, people have been predicting that technology will put them out of work since the Luddites, and it’s never come true. Instead, one of two things have happened. Either machines have augmented human workers, allowing them to produce more goods at lower prices, and so expanded industries so dramatically that overall they employ more people. Or displaced workers from one industry have gone into another – stable boys becoming car mechanics, or the like. There are a bunch of well-known theoretical mechanisms that compensate for technological displacement – see Vivarelli for a review. David Autor gives a vivid example:

Consider the surprising complementarities between information technology and employment in banking, specifically the experience with automated teller machines (ATMs) and bank tellers documented by Bessen (2015). ATMs were introduced in the 1970s, and their numbers in the US economy quadrupled from approximately 100,000 to 400,000 between 1995 and 2010. One might naturally assume that these machines had all but eliminated bank tellers in that interval. But US bank teller employment actually rose modestly from 500,000 to approximately 550,000 over the 30-year period from 1980 to 2010 (although given the growth in the labor force in this time interval, these numbers do imply that bank tellers declined as a share of overall US employment). With the growth of ATMs, what are all of these tellers doing?

Bessen observes that two forces worked in opposite directions. First, by reducing the cost of operating a bank branch, ATMs indirectly increased the demand for tellers: the number of tellers per branch fell by more than a third between 1988 and 2004, but the number of urban bank branches (also encouraged by a wave of bank deregulation allowing more branches) rose by more than 40 percent. Second, as the routine cash-handling tasks of bank tellers receded, information technology also enabled a broader range of bank personnel to become involved in “relationship banking.” Increasingly, banks recognized the value of tellers enabled by information technology, not primarily as checkout clerks, but as salespersons, forging relationships with customers and introducing them to additional bank services like credit cards, loans, and investment products.

This kind of thing has been remarkably consistent – so much so that arguments that today is the day technological unemployment happens should be treated with the same skepticism as arguments that today is the day we build a perpetual motion machine that works.

The argument in favor: look, imagine there’s a perfect android that can do everything humans do (including management) only better. And suppose it costs $10 to buy and $1/hour to operate. Surely every business owner would just buy those androids, and then all humans who wanted to earn more than $1/hour would be totally out of luck. There’s no conceivable way the androids would “augment” human labor and there’s no conceivable way the displaced humans could go into another industry. So at some point we’ve got to start getting technological unemployment. Here the vivid example comes from Gregory Clark

There was a type of employee at the beginning of the Industrial Revolution whose job and livelihood largely vanished in the early twentieth century. This was the horse. The population of working horses actually peaked in England long after the Industrial Revolution, in 1901, when 3.25 million were at work. Though they had been replaced by rail for long-distance haulage and by steam engines for driving machinery, they still plowed fields, hauled wagons and carriages short distances, pulled boats on the canals, toiled in the pits, and carried armies into battle. But the arrival of the internal combustion engine in the late nineteenth century rapidly displaced these workers, so that by 1924 there were fewer than two million. There was always a wage at which all these horses could have remained employed. But that wage was so low that it did not pay for their feed.

There may be some point at which we too stop being worth more than it costs to replace us. And the decline of manufacturing, the increase in labor force nonparticipation and despair in rural Rust Belt communities, etc, suggest that point is fast arriving.

This is a look at which of those arguments is right. Part I will investigate whether unemployment is getting worse. Part II will investigate whether that is because of technology. Part III will investigate what longer-term trends we should expect.

As usual, this is very long.

I.

Is unemployment actually getting worse?

Officially it’s at historic lows.

Source: Bureau of Labor Statistics. Just assume this for any graph that looks like this.

But the real concern isn’t about unemployment per se, but the labor force participation rate (from here on: LFPR). Unemployment measures how many people are looking for work; LFPR measures how many people are out of work and not looking. If people are so discouraged that they’ve given up looking for work, this would not show up in unemployment but would show up in LFPR. Here’s LFPR:

Remember, higher means more people working

What’s going on here?

Same data as before, only disaggregated by gender. From 1950 to 2000, workforce participation rises as women enter the workforce. But throughout this time, men are leaving at almost the same rate, leaving only a moderate net participation gain (and incidentally answering the question that confused me here). Around 2000, all the women who want to be in the workforce are there already, and the declining male trend takes over for a net decline. Because women’s increasing workforce participation is shaped by unrelated cultural trends, most of the rest of this article will focus on male LFPR (from here on: MLFPR).

A big fraction of declining MFLPR is Baby Boomers retiring. If most people are young, workforce participation will be high. If most people are old, workforce participation will be low. Economists adjust for this by taking something called prime age male labor force participation rate (from here on: PAMLFPR; if you think all these acronyms are getting annoying, I guarantee you it’s more annoying to read papers that just keep saying “prime age male labor force participation rate” again and again). Here it is:

Source for this and subsequent similar-looking graphs.

In the 1950s, ~97% (!) of prime-age men had a job. Today that number is more like 88%. This is the decline people are worrying about when they talk about technological unemployment or any other threat to work, and it seems to be happening across the Western world:

Frickin’ Germany, always making everyone else look bad.

The next few paragraphs are based on data from a Scott Winship’s Mercatus Center report on this. Their conclusion is contrarian, and they’re a libertarian-ish think tank which means they have some risk of bias. I’m citing them anyway because they have really fascinating data presented much better than anyone else, but keep this in mind.

Winship’s first point is that the decline in PAMLFPR doesn’t seem to be caused by people who can’t find jobs:

The first graph shows officially designated “discouraged workers”, people who were previously unemployed and looking for jobs but eventually gave up. The definition changed around 1990, but they never seem to be more than about 10% of prime age male labor force nonparticipators (from here: PAMLFNPers).

The second graph shows what percent of PAMLFNPers claimed to be looking for jobs, based on a survey that was only given until 1993. It shows only about 20% of them were interested.

The third graph is using slightly differently-parsed data to try to continue the trends after 1993. It’s sort of unfair, because it separates out “disability” into a separate category and assumes none of them want jobs. But about 30% of people on disability do say they want a job. It’s unclear exactly what they mean (are they just saying the wish they weren’t disabled? or that they might be willing to come off disability if a properly non-physically strenuous job became available?) but even if we count them, the percent of PAMLFNPers who want a job never goes above 30%.

If not discouraged workers who can’t find jobs, what’s going on? Here’s Winship’s answer:

Prime age non-working men are mostly on disability. But some are also in school (despite having to be above 25 to be included as “prime age”), retired (despite having to be below 55), or homemakers (remember, these are all men). Again, only about 1% (out of the total of 12%) say they can’t find work.

If we were very optimistic, we could paint a rosy picture of what’s going on here. The increase in disability represents improving social safety net that allows disabled people to be better supported. It’s great that more people are financially secure enough to retire early. It’s great that more people are pursuing a graduate education that has them in school after age 25. It’s great that gender stereotypes are decreasing and more men feel comfortable as homemakers, perhaps supported by a working spouse.

This is basically Winship’s account, although he is concerned that increasing disability benefits are discouraging work. He cites a bunch of papers to that effect which you can find in his footnote 42, and which aren’t super relevant to the question at hand.

But what would the pessimistic interpretation look like?

The next few graphs and some of the analysis below comes from Brookings Institute, another potentially biased think tank

In 1970, educated and uneducated men were about equally likely to be PAMLFNPers. The rate for educated men didn’t change. The rate for uneducated men shot up.

And I won’t show you graphs, but there are similar trends for poor people, ex-convicts, blue collar workers, and minorities. These are not the sort of people who are likely to be able to retire early, pursue graduate school, or defy gender norms. But they are the sort of people who might have trouble finding work. This is pretty suspicious. Also:

Labor force nonparticipation is increasing primarily in poor and lower-middle-class people without a lot of good options, just as their remaining options get much worse. Surely this suggests something worse is going on.

The easiest place for this to happen is disability. It doesn’t require disability fraud, per se. It just requires some people on the threshold of disability who are motivated by marginal cost/benefit analysis.

Suppose that you have bad back pain. You work in the auto factory, like your father and his father before him. Your back pain flares up pretty often, but you know your foreman pretty well and he gives you an easy shift until it passes, and the union makes sure that nobody gives you any grief about it. You like your company and your coworkers and you want to make them happy. Also, if you didn’t work, you would starve to death.

Now suppose that your factory closes, and the only job available is being a home health aide. This involves a lot of bending over and puts you in constant almost-unbearable pain. And it’s run by a giant faceless corporation which always seems to be trying to screw you over. Also, you live in West Virginia and are very manly, and changing diapers in nursing homes seems like undignified women’s work. Also, the pay is half what you’re used to. Also, the government just passed a new law making disability benefits much more generous and easier to get. So…

The graph shows that PAMLFNPers generally have terrible health. So real disabilities must have something to do with this. But Winship presents a lot of evidence that illnesses and chronic pain haven’t gotten worse over time, so it can’t fully explain the rise. The “gradually worsening job pushes person with serious disability over the edge” hypothesis has a lot going for it. Also: although 96% of people on disability say they’re out of work because of a health problem, 46% also say they’re out of work because there are no good jobs available (source).

The problem with this is: disability really doesn’t represent that much of the rise in PAMLFNP since 1960. Looking at our graph above, it goes from 2% of workers in ’68 to maybe 6% of workers now. And surely some large fraction of those people are actually disabled in ways that have nothing to do with their social circumstances. We’re talking like 2 percentage points tops.

Okay, fine. Let’s say you’re our West Virginia factory worker again, only now you can’t get on disability. Now what?

Maybe you choose to retire. And maybe you’re 53 years old and this isn’t the most reasonable financial plan, but you own your house, you get food stamps, and you can do odd jobs around your friends’ farm to make some extra money.

Or maybe you choose to go to that ridiculous Coal Miner To Coder school that got profiled on NPR a little while ago, in the hopes that you can have a pathway to a new career, or just so that you have something to do.

Or maybe you choose to stay at home with your kids, while your wife does the home health aide thing, and if anybody asks, you’re a “stay-at-home dad”.

And then when economists look at the statistics, they say “Oh, look, there’s no problem here, it’s just a combination of retirees, students, homemakers, and the disabled.”

I realize this is a stretch, especially since you would expect such a person, unless they were very self-deluded, to identify as “looking for work”. But the only sense I can make of all this is a model where, the more miserable your work is, and the more decent options you have available to you, the more likely you are to leave work.

If you’re very conservative, you might say – aha, I knew that people were just unemployed because they’re lazy!

But if you’re more progressive, you might ask – exactly how miserable do you have to be before you stop working? Should people with broken legs literally drag themselves on all fours to their workplace, just because it’s not physically impossible? I know that “I refuse to do this job because it’s too undignified for me, let me go on the public dole” doesn’t really win you a lot of social credit. But maybe conservatives could find it in their heart to be sympathetic to our hypothetical West Virginia factory worker with a bad back, who’s proud of having worked his whole life, but who feels like having to pivot at age 53 to changing diapers in nursing homes for minimum wage isn’t his cup of tea?

(remember, the other thing that’s way up among this same demographic is suicide – and probably for the same reason)

But even if we assume half the increase in disability plus a quarter of the increase in the other things is due to employment issues, employment issues still really only explain about 3 pp of the 10% increase since the 1960s. I can’t think of any reasonable assumptions where they explain more than half.

I like Derek Thompson’s discussion of this question, because he’s the only writer who seems to share my sense of puzzlement. There are all these men who seem miserable and who have vanished from the labor force. We all know it’s true. But the statistics don’t really seem to reflect or shed light on it. Somehow we, as a country, have managed to just lose several million working-age men. Maybe Donald Trump is going to look behind the White House couch one day, and find a large portion of the male population of the Southeast under the cushion. I don’t know.

In the next part, we’ll talk about whether automation explains the decline in labor force participation. But let’s keep in mind that the argument that there is a significant meaningful decline in labor force participation to explain, aside from people going to school and having more access to disability benefits and things like that – is not on super-solid ground.

II.

Recently, US manufacturing jobs collapsed. US manufacturing is still doing just fine in terms of number of widgets produced. It just no longer employs that many people.

Source 1, Source 2

The two graphs differ a bit in emphasis. The first, in raw numbers, makes it look like there was a big discontinuity around 2000. The second, in percent, suggests that percent manufacturing has been shrinking for a long time. This could represent a long-term decline in manufacturing – but more likely, it represents women entering the workforce; women are much less likely than men to have manufacturing jobs. The reality is probably somewhere in between the two.

Is the decline in manufacturing – whichever timescale we choose to look on – due to automation, or to other causes like globalization? A Ball State University report argues for the former:

Had we kept 2000-levels of productivity and applied them to 2010-levels of production, we would have required 20.9 million manufacturing workers. Instead, we employed only 12.1 million.

The report gets summarized in a few places as “13% of job loss is due to trade, 87% is due to increasing productivity/automation”, which seems like a fair summary of some of its claims. Although some commenters raise doubts, its numbers are in the same direction as the conclusions of economists Autor, Dorn, and Hanson, who published a series of papers finding that that “import competition explains one-quarter of the contemporaneous aggregate decline in US manufacturing employment”.

Hasn’t productivity actually been growing more slowly than usual recently? See for example this this Brookings paper, which notes that:

The past decade has seen slowdowns in measured labor productivity growth across a broad swath of developed economies. Aggregate labor productivity growth in the U.S. averaged 1.3% per year from 2005 to 2015, less than half of the 2.8% average annual growth rate it sustained over 1995 – 2004. Similarly sized decelerations were observed between these two periods in 28 of 29 other countries for which the OECD has compiled productivity growth data

It continues:

The drops in productivity growth have struck some as paradoxical, given the seemingly brisk pace of technological progress and plethora of new products that have been introduced and diffused throughout the world during the slowdown period. Indeed, many have suggested that the slowdown is substantially illusory, a figment of the inability of current economic statistics to capture the true rate of technological advance in standard productivity metrics.

If the productivity slowdown were illusory, that would help explain the apparent speedup in automation, and all those job losses. Sounds promising, but what do all the economists in the world think?

Darn.

Justin Fox has some more in-depth analysis here and also concludes productivity is not that great.

Are we allowed to say “that’s just how things work”? Like, agricultural productivity increased for millennia, but didn’t lead people to abandon agriculture en masse until the Industrial Revolution – when it did exactly that. In 1790, 90% of Americans were farmers, even though agricultural productivity had been improving for ages. Today, 2.6% of Americans are. Maybe manufacturing just had the same kind of moment. Advances in technology can put farmers out of work (but shift them to manufacturing) – surely it can put manufacturers out of work (and shift them to ???).

So, if 70% – 80% of manufacturing job losses were due to automation, might automation be responsible for the decline in PAMLFPR, thus revealing the elusive technological unemployment?

The time trend in absolute number of manufacturing jobs reveals a sharp drop around 2000, which doesn’t match change in PAMLFNP. The trend in percent reveals a gradual decline, which does sort of match the PAMLFNP trend, but because the percent is likely to indicate women, who are not included in PAMFLNP, it might not be as significant. But there are two other reasons why these don’t seem to match very well.

First, disability represents the main route by which people could plausibly be leaving the labor force to become PAMLFNPers. Here’s the graph:

There is a general trend of increasing disability since 1985, but a paper by Autor and Duggan suggests this is almost entirely due to a reform of the disability system around 1984 which made it easier to get benefits (also, the size of the workforce increased due to more female participation, meaning the pool of potential disabled people increased too).

Second, it is hard to explain why now. Whether we’re worrying about the past twenty years of automation (collapse in raw numbers of manufacturing jobs) or the past sixty (decline in PAMLFPR, decline in percent manufacturing) – neither corresponds to the several hundred years in which things have been getting more automated. Why didn’t previous eras of improving automation result in job loss?

Let’s go back to all the economists in the world and see what they think:

Economists very strongly believe automation has not historically reduced employment. But they do believe automation is making wages stagnate right now. I don’t really understand what’s going on here. Are they saying that automation can depress wages, but not reduce employment? Surely (given the existence of a minimum wage) that doesn’t make sense. Or are they saying that automation never caused any problems before, but it is causing problems now?

The site offers some of the economists the chance to explain what they meant, and a lot of them seem to be saying that automation has temporarily caused problems in the past, but they always resolved with time as new industries open up. Maybe we’re just in a temporary bad period? Likewise, one economist who agrees that automation caused wage stagnation says that “it may have a short-run impact but there is no reason to believe that it is permanent.”

All of this is a mess. But the impression I get from this mess is that there is little sign of technological unemployment happening today in a historically unique way, or even picking up pace. I get this from a few sources.

First, the official unemployment rate looks great, so if we were going to make this argument we would have to do it off of PAMLFPR.

Second, Winship’s optimistic take on PAMLFPR is hard to easily refute. PAMLFNPers pretty clearly say they’re not looking for jobs, and they’re just perfectly innocuous students, retirees, etc. We have trouble believing them, especially based on their demographics. But it’s very hard to look at the increase and see a place where unemployment issues could have slipped in.

Third, PAMLFPR has been getting worse gradually since about 1960, with no sign of any recent worsening. It is hard to explain why technological unemployment would have started around that time – at least if we limit our explanations to the nature of technology alone. And it doesn’t seem to match the more sudden decline in manufacturing around 2000.

Fourth, most economists appear to remain doubtful of the possibility of long-term technological unemployment.

I realize this goes against common sense. Maybe I’m missing something and totally wrong here. But if I am forced to interpret the data as I see it, I just don’t see the signs of technological unemployment. It’s just not there.

And in my defense, this also seems to be the opinion of David Autor, the main economic expert on this subject. In an interview with The Economist, he said that there was “‘zero evidence’ that AI is having a new and significantly different impact on employment”.

III.

This doesn’t mean everything is great. As the IGM panel shows, even if robots aren’t putting people out of work, they may be causing wages to stagnate. The people getting kicked out of manufacturing jobs may have other jobs available to them (and so not end up affecting the PAMLFPR numbers), but those jobs may not be as good or pay as well. This isn’t “technological unemployment”. But it might be technological underemployment.

Most people expect that technological unemployment will hit the least skilled first, but that doesn’t seem to be entirely true. This chart and some of the following analysis are going to be from the Heritage Association – another potentially biased think-tank, but hopefully without much reason to obfuscate these issues.

The best-paying jobs – managers, professionals, and the like – are doing fine. The lowest-paying jobs, like personal care and food, are also doing fine. It’s the middle-paying jobs that are in trouble. Some of these are manufacturing, but there are also office and administrative positions in the same categories.

This is potentially consistent with a story where the jobs that have been easiest to automate are middle-class-ish. Some jobs require extremely basic human talents that machines can’t yet match – like a delivery person’s ability to climb stairs. Others require extremely arcane human talents likewise beyond machine abilities – like a scientist discovering new theories of physics. The stuff in between – proofreading, translating, records-keeping, metalworking, truck driving, welding – is more in danger. As these get automated away, workers – in accord with the theory – migrate to the unautomatable jobs. Since they might not have the skills or training to do the unautomatable upper class jobs, they end up in the unautomatable lower-class ones. There’s nothing in economic orthodoxy that says this can’t happen.

David Autor and his giant block of citations agree:

Because jobs that are intensive in either abstract or manual tasks are generally found at opposite ends of the occupational skill spectrum—in professional, managerial, and technical occupations on the one hand, and in service and laborer occupations on the other—this reasoning implies that computerization of “routine” job tasks may lead to the simultaneous growth of high-education, high-wage jobs at one end and low-education, low-wage jobs at the other end, both at the expense of middle-wage, middle education jobs—a phenomenon that Goos and Manning (2003) called “job polarization.” A large body of US and international evidence confirms the presence of employment polarization at the level of industries, localities, and national labor markets (Autor, Katz, and Kearney 2006, 2008; Goos and Manning 2007; Autor and Dorn 2013; Michaels, Natraj, and Van Reenen 2014; Goos, Manning, and Salomons 2014; Graetz and Michaels 2015; Autor, Dorn, and
Hanson 2015)

The fall of middle-skill-level jobs has led to a corresponding fall in middle-income jobs:

Note that, contrary to an extremely pessimistic picture, this would suggest that most people who leave middle-paying jobs go to higher-paying jobs

And with a corresponding decline in the fortunes of the middle class:

Note that this does not really back up the optimistic picture from above.

Why is this happening now when technological progress has been going on forever? This gets into the whole decline-of-the-middle-class argument, a giant political morass featuring de-unionization, regulation, automation, globalization, the 1%, and pretty much everything else. Is there also a role for today’s robots just plain being better than yesterday’s Rolodexes and whatever else the forefront of technology was? Or our education system being less able to cope with them? I’m not sure.

As far as I know, there is no economic theory stating that the number (or percent) of high-skilled jobs must always stay the same. I’m also not sure how to include fixed cognitive skills (eg some people are smarter than others) in this question. An optimist might argue that things will get better as today’s obsoletely-trained workforce retires and tomorrow’s trained-for-the-appropriate-jobs workforce graduates. But maybe this is better viewed as a race between two competing forces; generational churn producing students with the right set of skills, and technology making new skills obsolete. I don’t know why this should have increased recently, but it seems like – at least for the middle class – this is a race they are now losing.

IV.

Predicting the future is naturally harder than observing the present, since we have data about the present and not the future. But the data about the present is contradictory and incomprehensible and just makes things more confusing, so we might as well try going with the future and seeing how we do there.

We’ll start with those surveys of economists again, since they seem like the people most likely to know. Here’s a panel of top European economists on the future of technological unemployment:

In the same survey, 93% of economists with an opinion on the issue agreed that the economic benefits of robots will be so great that they could be used to compensate the workers who were negatively effected. But in a survey I conducted in my imagination, 100% of people who have not been living in a cave the past two hundred years agreed that this will never happen in real life.

So economists really have no idea about any of this. What are we paying them for, anyway?

Frey and Osbourne analyze what jobs are most susceptible to automation. They claim that “47% of total US employment is at risk”. This sounds suspiciously precise and it’s unclear their numbers have any relationship to reality. They do find “evidence that wages and educational attainment exhibit a strong negative relationship with an occupation’s probability of computerisation”.

Overall none of this seems to be making things much clearer.

I’ve cited David Autor something like five times already. He is the recognized expert in this area – I blame nominative determinism – and has written widely. His own opinion is:

I expect that a significant stratum of middle-skill jobs combining specific vocational skills with foundational middle-skills levels of literacy, numeracy, adaptability, problem solving, and common sense will persist in coming decades. My conjecture is that many of the tasks currently bundled into these jobs cannot readily be unbundled—with machines performing the middle-skill tasks and workers performing only a low-skill residual—without a substantial drop in quality. This argument suggests that many of the middle-skill jobs that persist in the future will combine routine technical tasks with the set of nonroutine tasks in which workers hold comparative advantage: interpersonal interaction, flexibility, adaptability, and problem solving. In general, these same demands for interaction frequently privilege face-to-face interactions over remote performance, meaning that these same middle-skill occupations may have relatively low susceptibility to offshoring. Lawrence Katz memorably titles workers who virtuously combine technical and interpersonal tasks as “the new artisans” (see Friedman 2010), and Holzer (2015) documents that “new middle skill jobs” are in fact growing rapidly, even as traditional production and clerical occupations contract.

This prediction has one obvious catch: the ability of the US education and job training system (both public and private) to produce the kinds of workers who will thrive in these middle-skill jobs of the future can be called into question. In this and other ways, the issue is not that middle-class workers are doomed by automation and technology, but instead that human capital investment must be at the heart of any long-term strategy for producing skills that are complemented by rather than substituted for by technological change. In 1900, the typical young, native-born American had only a common school education, about the equivalent of sixth to eighth grades. By the late 19th century, many Americans recognized that this level of schooling was inadequate: farm employment was declining, industry was rising, and their children would need additional skills to earn a living. The United States responded to this challenge over the first four decades of the 20th century by becoming the first nation in the world to deliver universal high school education to its citizens (Goldin and Katz 2008). Tellingly, the high school movement was led by the farm states. Societal adjustments to earlier waves of technological advancement were neither rapid, automatic, nor cheap. But they did pay off handsomely.

Do we really have evidence that compulsory schooling was a result of increasing automation? If so, could we tell a story where the gradually increasing length of schooling – from minimal, to primary school, to high school, to “you better get a college degree or you’ll regret it later”, to increasing pressure to go to graduate school – is a reaction to automation and the threat of technological unemployment? Could this be the reason why automation finally seems to be causing problems – a financial and cultural inability to extend schooling any further than it’s already gone?

Or – inspired by Bryan Caplan’s The Case Against Education – could we tell the opposite story? One where increasing credentialism makes it harder for people whose jobs have been automated to switch careers the way they did before? Higher-paying jobs no longer just require skills, they require a college degree in a relevant field – which is very hard for a mature worker to get. Being an office manager and being a nurse are both middle-income jobs – but in the past, an office worker would have needed about six months of inexpensive training plus a lot of on-the-job apprenticeship to become a nurse, whereas now they would need a four-year Bachelors of Science in Nursing from a university whose price tag has dectupled for no reason over the last half-century. A now-unemployed office manager might have been able to afford the first even if middle-class; the second might be well beyond her reach. Unable to shift into another middle-class job, she is forced to take a lower-class job as a fast food worker or something.

I am not entirely sure how differences in cognitive ability fit in here. My guess is to a first approximation they don’t – if standard economic theory is correct, it should be possible to create middle-paying jobs that use the full potential of people with any amount of cognitive ability, taking advantage of various human cognitive skills that are difficult to automate. Although some naive takes like “everyone should just become a programmer” fail to understand this, I don’t think the entire argument is based on misunderstanding of this point, or that it forms a particularly strong counterargument.

Anyway, if Autor’s prediction is “we will be able to weather this danger as long as our education system is able to rise to meet the challenge”, I’m just going to round this off to “we’re super doomed”. But I think his methodology – of noticing that we always met the challenge before, and trying to figure out what might be different this time – is a promising one.

Finally: we’ve been talking about economists a lot here, but what about the roboticists? Aren’t they relevant too? Grace et al survey top AI researchers on when AI might be able to replace humans in various things; these researchers don’t necessarily know anything about economics, but they at least know something about progress in robotics. On average, they believe AI will reach human performance at truck driving, retail selling, translation, transcription, and bipedal running all before 2030. Whether those robots will be affordable, widely adopted, or able to deal with the long tail of real-world situations like blackouts or vandals or bad weather – is a different story. The point is, roboticists are pretty sure they’ll have their contribution to the economic takeover ready pretty soon.

(they do say robots won’t be writing bestselling novels until 2050, so JK Rowling’s job is safe for now)

They also say that robots will be able to do all human tasks, including novel-writing, science, and further AI research – sometime between 2050 and 2150. At that point, obviously, all bets are off, and we have a lot more than unemployment to worry about.

V.

Here are some tentative conclusions:

1. Technological unemployment is not happening right now, at least not more so than previous eras. The official statistics are confusing, but they show no signs of increases in this phenomenon. (70% confidence)

2. On the other hand, there are signs of technological underemployment – robots taking middle-skill jobs and then pushing people into other jobs. Although some people will be “pushed” into higher-skill jobs, many will be pushed into lower-skill jobs. This seems to be what happened to the manufacturing industry recently. (70% confidence)

3. This sort of thing has been happening for centuries and in theory everyone should eventually adjust, but there are some signs that they aren’t. This may have as much to do with changes to the educational, political, and economic system as with the nature of robots per se. (60% confidence)

4. Economists are genuinely divided on how this is going to end up, and whether this will just be a temporary blip while people develop new skills, or the new normal. (~100% confidence)

5. Technology seems poised to disrupt lots of new industries very soon, and could replace humans entirely sometime within the next hundred years. (???)

This is a very depressing conclusion. If technology didn’t cause problems, that would be great. If technology made lots of people unemployed, that would be hard to miss, and the government might eventually be willing to subsidize something like a universal basic income. But we won’t get that. We’ll just get people being pushed into worse and worse jobs, in a way that does not inspire widespread sympathy or collective action. The prospect of educational, social, or political intervention remains murky.

523 Responses to Technological Unemployment: Much More Than You Wanted To Know

  1. Calion says:

    I’ve not made my way through this yet; instead I got distracted by the “Invisible Women” post you linked, in particular this sentence: “They do seem to have stagnated dramatically starting 1970 or so, but I feel like the income inequality explanation for that is on pretty solid ground.”

    Do you still feel that way? Because that is not my understanding at all. My understanding is that wages *including non-wage compensation* did not at all stagnate in the 1970s, but continued more-or-less apace. Health insurance became part of the compensation package for most workers, and so wages had to decline in order to offset the increase in benefits.

    • Scott Alexander says:

      Vague memory of having heard this claim, found it plausible, looked into it, found it wasn’t true. Can’t give you more than that. A fair look into that would probably be as long as the original post.

      • Swami says:

        Scott,

        Most estimates, when adjusted for transfers and taxes, proper inflation, household size, and non monetary benefits reveal that median incomes have increased somewhere in the neighborhood of 40 to 60%. This however still grossly understates welfare gains to real families as it excludes movement up to higher classes as people age and it ignores that the average is being weighted down by the influx of 40 million immigrants over the last 30 years or so.

        Here are references:

        The first is from the Minneapolis FED. Follow the link for details.

        https://www.minneapolisfed.org/publications/the-region/where-has-all-the-income-gone

        “The main finding is that—after adjusting the Census Bureau data for three key factors—inflation-adjusted median household income for most household types increased by roughly 44 percent to 62 percent from 1976 to 2006.

        Here is a preview of the key data issues that lead to the higher estimates of median household income growth.
        1. The price index used by the Census Bureau overstates inflation, and thus understates income gains, relative to a preferred price index. 

        2. A changing mix of household types leads the overall median increase to understate the median increase of most household types.

        3. The Census Bureau measure of household income understates income growth by excluding some rapidly growing sources of income.”

        For additional elaboration here are some other good links which explain the income gains…

        https://www.cnbc.com/2014/01/29/-wage-stagnationcommentary.html

        https://ideas.repec.org/p/nbr/nberwo/19110.html

        And remember, most of these still don’t account for the effects of immigration, which totally distorts real gains for real families (specifically including those immigrating, who while personally benefitting have the effect of lowering the US median by crossing the border).

        • ADifferentAnonymous says:

          “Transfers and taxes” are not part of “wages”. If we’re interested in inequality of material well-being, then we care about transfers and taxes–but for a topic like technological un(der)employment, we care about what people’s labor is worth on the market.

          (Whereas non-monetary benefits like employer-based health insurance do count towards this)

          • Swami says:

            Yes they are two distinct issues. Thanks for adding the clarification.

            But, one issue which we are trying to explain is labor rate participation, which is a factor of net benefit of working vs benefits of not working, at least in the formal market. Transfers and taxes are extremely important to the conversation and tradeoffs.

            I certainly agree that transfers to non workers (men or their wives/girlfriends) is contributing to the lower participation rate of males.

            In summary, individual living standards have not stagnated anywhere as much as advertised (they are the highest ever, and up 40 to 60% from 40 years ago), median wages including benefits adjusted for inflation are better than ever (but not rising at as fast a rate), and the incentives to not enter the workforce for a sub segment of the population are higher than ever for reasons explained in my other comment.

          • Brad says:

            “Transfers and taxes” are not part of “wages”. If we’re interested in inequality of material well-being, then we care about transfers and taxes–but for a topic like technological un(der)employment, we care about what people’s labor is worth on the market.

            I agree as to transfers, but not as to taxes. What people’s labor is worth on the market is asking about employers’ willingness to pay, not what is received by the employee. If payroll taxes one day increase by 3% of payroll, and the incidence ends up being born entirely by the employees, then total compensation will have dropped, but nothing will have changed in terms of what people’s labor is worth on the market. The same labor is worth the same amount to employers, and they are still paying that same amount, it’s just that more of it is going to the government and less to the employee.

          • Antistotle says:

            Of course they are part of wages.

            They are based on a percentage of my wages and subtracted from my wages when I get paid.

      • Freddie deBoer says:

        Don’t listen to these libertarian cranks, Scott. The large majority of academic economists agree that wages have stagnated in just the way commonly understood.

        • Swami says:

          Why would you call someone a libertarian or a crank, Freddie?

          I am sure I am not the former, and would gladly refute the latter with actual data absent ad hominem attacks. If there is an error in the Minn FED report, please enlighten us.

          Median income is up forty to sixty percent. If we adjust for thirty to forty million immigrants, the numbers go up even more (for non immigrants by about eight to ten percent if memory serves), and much more for the immigrants (who move from a low income economy to a high income economy).

          • Freddie deBoer says:

            You say that as though you are unaware that the overwhelming consensus of established economists disagree with your conclusion.

          • moonfirestorm says:

            @Freddie:

            Could you please expand on this? Swami’s brought up some interesting points, and “the consensus says” isn’t really much of a rebuttal.

            What do you think Swami is missing here?

          • Swami says:

            Freddie,

            I challenge you to find any economist who disagrees with the Minneapolis Fed paper. We can go on from there.

            I am well aware of conventional wisdom repeated as mantra with the hope that it drowns our reason and analytical rigor. If you repeat a half truth often and loudly enough a closed group can convince themselves of just about anything. Add ad hominem attacks to those actually noting the emperor is not wearing any clothes, and you can lock in any belief.

            Let me be real clear, Freddie. I offer data and links and can supply gobs of others all showing the same thing — incomes have not stagnated as much as advertised (up at least 40% before accounting for immigration). The rational response on your part would be to take on my arguments and data, not to call me names, label me unfairly, and resort to conventional wisdom.

        • herculesorion says:

          This is interesting because it lines up with something in the OP: That economic statistics seem to be deranged because we’re using the wrong statistics. The calculation of “real wages” is based on goods and services available in the early 1970s. We barely even live on the same planet, technologically, as the baseline for that calculation.

          • Swami says:

            I would argue that most of the wealth of the past few decades is completely outside of traditional economic measures. Most of the new is damn near free on the margin. I could add a couple trillion dollars to GDP just by adding in email, photos, video conferencing, GPS devices, reading and video material, music streamers and such available at no marginal cost on our phones.

            Economists are terrible at measuring free, user produced and non rivalrous goods. I think what they measure as negative GDP is actually massively positive production which is disappearing from their radars. They also do a terrible job at measuring improvements in quality over time.

          • Antistotle says:

            In 1983 the cheapest car on the market (from memory) was a light truck with automatic transmission, hand cranked windows and an am-fm stereo. It was advertised (again from memory) at about 9k.

            Today the bottom of the barrel…er, cheapest new car is about 13k if you can find one. It’s still got a manual transmission, but usually has air conditioning, power windows and an AM/FM/MP3 stereo.

            In 1990 a CD about 16 bucks. Today you can stream music 24×7 for about the cost of 6 CDs per year. Well, you can do it free if you’re willing to put up with ads.

            In 1990 a loaf of wonder bread cost who f*king cares because no one but the poorest of the poor buy that s*t anymore. Whole wheat high protein cruelty and gluten free BABY!

        • christhenottopher says:

          I haven’t been able to find a survey of economists on exactly the question of wage stagnation, but I did find a survey on the question of whether “The 9% cumulative increase in real US median household income since 1980 substantially understates how much better off people in the median American household are now economically, compared with 35 years ago.” The result? 60% strongly agree or agree with the above statement, 21% uncertain, 10% disagree (no strongly disagrees). Weighted by confidence, 70% agree. If anything the consensus of economists in that survey would seem to be directly against the idea that “wages have stagnated in just the way commonly understood.”

          • Swami says:

            Thanks Christhenottopher,

            I will take that as an strong economists’ consensus that wages have not stagnated.

            Still hoping to actually find out if Freddie has anything to add constructively to the issue.

            I have been researching the issue for a while now, and have come to the conclusion that;
            1). Wages have gone up much more than conventionally represented and
            2). There is a zeitgeist being promoted of exaggerated stagnation.

            I certainly don’t think everything is rosey. Growth could have been even better in developed nations (it certainly is the best in the history of humanity globally over past twenty years, but that contradicts the zeitgeist of the age too).

      • chosh says:

        This is where I most recently saw that claim. Could definitely believe it doesn’t explain the full gap, but would need some pretty strong evidence to be convinced it isn’t a factor at all.
        https://sadoeconomist.tumblr.com/post/158460248954/do-you-have-some-convenient-source-numbers-for-the
        Edited to include later comments in the discussion.

        • AG says:

          http://jonathan-tepper.com/why-american-workers-arent-getting-a-raise-an-economic-detective-story/

          Some economists have argued that the gap between wages and productivity is an illusion. They argue that much of the gap can be explained by year-end bonuses, which are not included in hourly pay, by healthcare costs, which doesn’t show up in a paycheck but the worker benefits from, and by stock options, which also doesn’t show up in a paycheck. However, we can discount these explanations. Healthcare, bonuses and options are a real expense to companies, and if companies were getting hit with these costs instead of wages, it would show up in corporate profit margins. Today, corporate profit margins would not be at record highs. If the divergence between wages and productivity is real, the difference should clearly shows up in corporate profits, and it does.

          • The Nybbler says:

            That’s just comparing two numbers that are not so simply related. We can in principle calculate total compensation for employees, but we can’t validly just say “profits are at record highs so employers aren’t paying their employees more”. Whether or not profits are indeed at record highs.

          • Cliff says:

            Profits are actually not historically high. Another case where most analysis is poorly done.

          • AG says:

            @Cliff:
            The quote was merely an excerpt. The article itself has a more specific category of profit they’re looking at.

            Corporate profits as percentage of Gross Domestic Profit (GDP) are near record highs and labor’s share of GDP is near record lows. The trend in corporate profits is a mystery to economists and investment strategists. Jeremy Grantham, a well-known investor, has pointed out, “Profits are the most mean reverting series in finance. If margins don’t revert something has gone wrong with capitalism.” Employee Compensation as a Percentage of GDP has been falling for years.

            And these profit numbers don’t even get into the stock buyback money that doesn’t show up as profit: https://www.theatlantic.com/politics/archive/2015/02/kill-stock-buyback-to-save-the-american-economy/385259/

            Meanwhile: http://www.epi.org/publication/charting-wage-stagnation/

            Employers are cutting health care for young workers, both college and high school graduates

            So where is that mysterious other-benefits non-wage money again?

          • Ketil says:

            And these profit numbers don’t even get into the stock buyback money that doesn’t show up as profit: https://www.theatlantic.com/politics/archive/2015/02/kill-stock-buyback-to-save-the-american-economy/385259/

            Can somebody explain why that makes any sense at all? The linked article lists stagnant wages and budget deficit, and under-financed fire departments. And then:

            Where did all this money go?

            The answer is as simple as it is surprising: Much of it went to stock buybacks—more than $6.9 trillion of them since 2004

            As far as I understand, a company can either use its surplus money to pay dividends (cash) to shareholder, or it can buy back own stock from shareholders, inflating the share price. The Atlantic just makes no sense to me, but I also fail to see the link to profit hiding. Are stock buybacks treated as (tax-deductible) expenses for the company, or something?

      • Notsocrazy 24 says:

        I’ll put in my vote for “make this a future post” then.

  2. Bugmaster says:

    The argument in favor: look, imagine there’s a perfect android that can do everything humans do (including management) only better. And suppose it costs $10 to buy and $1/hour to operate. Surely every business owner would just buy those androids, and then all humans who wanted to earn more than $1/hour would be totally out of luck.

    Yes, but since this pretty much describes a post-scarcity society, what’s the problem ?

    • Calion says:

      I think the question here is, “who gets that lack of scarcity”? In a capitalist economy, that doesn’t necessarily automatically flow evenly to everyone.

      • Bugmaster says:

        Why not ? If robots are able to out-perform humans at pretty much every task, including finance; then what is the capitalist economy based on ? It can’t be money, because money is essentially worthless now that the robots run the entire financial industry. It could be some sort of rare natural resource, like gold or uranium, but what would you buy with your uranium coins, and who would want them ?

        • Scott Alexander says:

          It sounds like it would be based on access to the products of robots, and that this would be controlled by owners of robots. Surely robot-owners would sell their robots’ products for money, and use that to buy the products of other robots.

          • Bugmaster says:

            Ok, so basically you’re saying there’d be artificial scarcity, as robot-owners restrict the number of robots and/or throttle their output. But then, it seems like humans might be able to pick up the slack in manufacturing…

          • Scott Alexander says:

            No, why would this be artificial scarcity? Robots can only produce a finite amount just as humans can only produce a finite amount. If the robots were good it might be much more, but it just means rich people will pinch pennies to buy mile-long yachts instead of 500-foot-yachts.

          • mario rossi says:

            Are you assuming some kind of artifical limit on the number of robots? If there is no limit on the numbers of robots, then the hourly cost must represent the cost of the resources needed to build and operate the robot over its economic life. The competition between robot producers will push profits in the robot building sector to zero and there will be no surplus. No extra profit for robot owners or builders there… Why would robot ownership be concentrated? You are talking about businesses with explosive growth. Just save enough for one robot and that robot can build you a robot army…

            You could still argue that energy and raw materials would be scarse, but an infinite robot slave population makes almost any infrastructure project viable. Any renewable energy source is viable, any recycling project is likely viable. Any dangerous deposit can be accessed. Educated people require massive investments to train. A robot engineer would drop the cost of any construction project massively. Robot doctors would drop medical costs (and research) to almost zero. With a basically infinite almost free workforce terraforming Mars and Venus become possible. You could build space habitats with von neumann machines… The entire concept of scarcity becomes difficult to envisage.

            There is only one scenario in which scarcity would keep existing: a massive demographic explosion. Do we really believe that from current levels? There seems to be a pretty strong relationship between income level and wealth: the massive increase in wealth generated by the increase productivity is unlikely to push fertility higher. It’s not impossible, but it would be a large reversal of current trends.

            I think the only reason why people fear such scenario is not economic. Such a technology would enable dystopian society: there would be little need for society itself as you could imagine complete independence as long as you had enough land to fuel your robots (I guess Asimov imagined such society in Solaria). Robot armies could kill or enslave the human population.

            But even this seems fairly unlikely to me. Why would people bother? Most things we consider luxuries nowadays (education, medical care, live entratainment) would be basically free… What would you gain by killing or enslaving the rest of the human population? I am not saying someone wouldn’t try, but would a large percentage of the plutocrats really go along with it? I doubt it. Even if a few isolate themselves in their paradise enclaves, the increased productivity would still make everybody else unimmaginably rich.

          • bessiambre says:

            For most people, “working” in a world where machines are better than humans at most tasks would consist of going online, buying or trading an amount of energy, buying raw materials or spent matter that is ready to be recycled and telling your robots what you want. The robots would produce goods or services.

            Remember since robots would be built by robots, they would be relatively cheap, an obvious investment for anyone. They’re likely to be one of the first thing anyone would buy if they didn’t have one. Information copying is already really cheap. I assume there would be open source robot designs out there. You would only have to own the energy and raw materials to get robots to built other robots for you.

            The limiting factor for production would no longer be human labor but natural resources, energy and land because the machines would still required time, space, energy and matter to produce goods and services.

            Some people may work on designing new better machines that produce finer goods. This would be mostly creative work as the technical part would mostly be automated.

            People wouldn’t have to leave their dwelling to work. Machine owners could watch feeds of their machines working in an industrial park somewhere. The finished goods, spent matter (trash) and the machines themselves, would be picked up and delivered by self driving delivery robots.

            To get some variety, people might trade the production of different robots and they would trade excess spent matter. They would also trade the machine designs and the land or space to host the machines. The machines would sometimes have to be replaced when worn out or obsolete.

            The economic crux, would be inequality of ownership of energy production, and natural resources. Society might be constrained by the total energy production based on what can reasonably be captured from the sun. Hopefully governments wouldn’t allow ownership of fundamental physical resources to be too concentrated. A good way to prevent too high inequality would be for everyone to be shareholders in global energy production. Every day, shareholders would receive a dividend, an amount of energy to be spent. They could use it in machines to produce stuff or services, trade it for natural resources or even store it in a battery.

            In the best case scenario, if the technology allows us to become an “interplanetary species” some of the natural resource constraints problems might be alleviated.

          • Murphy says:

            “Just save enough for one robot”

            This hits the same problem that much of the third world faces re: certain diseases.

            If you have close to zero money with nothing else they want from you, you are worthless to a capitalist economy.

            And even if you manage to find something someone wants from you and scrape together the cash for a bot… your bot is now in competition with later models maintained by bigger and better organized setups. Even your bot is at a disadvantage vs the people who own a billion bots.

            Is there anything really stopping the situation from Manna where all the zero-asset citizens with no saleable skills just get shoved into a few hundred cubic miles of subsistence block housing and kept from trespassing on the real humans wealthy job creators land next door by bots in such a manner that the tiny fraction of the population who own the bots can simply forget you and everyone like you exist?

          • albatross11 says:

            There’s a really important idea here that is critical to the whole discussion. Lots of old-style class conflict was based on some people being exploited–I need your labor to make me rich, and I’m going to take measures to make you work for me for not very much money and under conditions maximally convenient to me.

            Assuming I’m not allowed to actually have slaves, the outcome depends on the market (are there many hungry people per job, or many urgently-needed jobs per capable person) and on what kind of power I have (can I get the national guard called out to break strikes?) But you have some hope of getting a decent outcome out of this situation, because you have some bargaining power–I actually need your labor. You might plausibly end up as a factory worker or coder or whatever with a decent middle-class standard of living.

            The modern/near-future class conflict seems like its likely to be very different. I don’t want to exploit you, I simply want nothing to do with you. Think of urban ghettos now–few companies want to set up factories there. Wages would be pretty low, but the companies don’t think they’d be low enough to compensate for the low quality of employees and the costs of doing business in those places. Even a lot of stores don’t want to do business in some of those neighborhoods–too little money + too much crime = food deserts.

            Exploitation might be bad (though it’s hard to define exactly what exploitation is), but uselessness is worse. And that’s what we’re ultimately worrying about–do we see an increasing fraction of people become useless to the economy–nobody wants to hire them because they’re not productive enough or they’re too much trouble.

        • Grek says:

          Control over the robots. If you start out rich and can buy a lot of robots when money still means something, you (and later generations of your family) end up with lots of robot slaves to tend to your every need (including the harvesting natural resources and the assembly even more robot slaves), while everyone else who couldn’t afford to buy robots during the transitional economy gets nothing forevermore.

          • Tracy W says:

            But if you have robot slaves producing everything you want, what’s left? The adulation of your fellow humans – philanthropy.

          • onyomi says:

            Money will always mean something so long as there is scarcity–for anybody.

          • Barely matters says:

            Buying yet larger and more extravagant ultrayachts to one up other robot owner overlords, much like what happens with billionaires today who can functionally buy anything they want?

            Second answer: Relative Positional goods. There can still only be a single gold medalist in a given area (Or one largest, coolest, most impressive yacht in the world) and there is still only a finite amount of property. Some things don’t scale with post scarcity.

          • Mark Bahner says:

            “…while everyone else who couldn’t afford to buy robots during the transitional economy gets nothing forevermore.”

            Until everyone getting nothing forever taxes away 90% of your income, and everyone lives happily ever after. 😉

          • Doesntliketocomment says:

            How are they going to be compelled to pay these punitive taxes? Remember, they have armies of robots.

          • Mark Bahner says:

            How are they going to be compelled to pay these punitive taxes? Remember, they have armies of robots.

            Yes, so they basically can keep billions or trillions of dollars, as long as they’re willing to kill all the people who have nothing.

            So there are two requirements:

            1) The wealth needs to be concentrated to an extent far greater than any extent previously observed on earth to date, and

            2) The people who have that wealth need to be pretty inhuman.

            I think a much, much more realistic scenario about which I’m truly worried is that the robots themselves have the power. Then they *are* literally inhuman. And: 1) Potentially essentially infinitely smarter than we are, and 2) Not requiring of most of the things humans need to survive, e.g. clean air to breath, clean water to drink, shelter from the elements, sleep, etc. etc. (As the Sting song goes, “How fragile we are, how fragile we are…”)

          • Civilis says:

            Second answer: Relative Positional goods. There can still only be a single gold medalist in a given area (Or one largest, coolest, most impressive yacht in the world) and there is still only a finite amount of property. Some things don’t scale with post scarcity.

            One of those relative positional goods in the post scarcity economy of Neal Stephenson’s The Diamond Age is physical goods such as furniture that are made by human hands. The more robots make furniture, the more scarce (and valuable as positional goods) human made furniture will be. So the robot owners will end up with human-made furniture, and the human furniture makers will end up with robot-made furniture. It might no longer be economically viable to have humans mass-produce furniture, but positional goods don’t need to be economically viable.

            Another point: the most valuable intellectual development in a near-post-scarcity situation is going to be to find a use for all that unused human labor sitting around doing nothing. Even if human labor is universally less productive than robot labor, an economy using human and robot labor is going to produce more than one relying solely on robot labor. Further, comparative advantage comes into play. If a robot miner produces 100x the output of a human miner, and a robot poet produces 2x the output of a human poet, economically your most productive economy is to make all robots miners and let the humans stick to writing poetry.

            If nothing else, there’s always the solution of building a virtual economy that includes scarcity inside your post-scarcity economy, especially if working in that virtual economy is more entertainment than labor. And this is something we already have and can see the mechanisms working. Right now, we have people trading the virtual currency in MMO video games earned from playing the game for hard currency.

          • Murphy says:

            @Mark Bahner

            Killing is probably not even terribly necessary.

            Historically many countries have set property requirements to have the vote at all. Or stripped the francise from people who committed minor crimes, people with various health problems etc.

            Stripping people on welfare of the vote would have lots of people clapping for it in our current world, prime it enough and you could probably shove it through.

            Then there’s no need for murder, indeed the incredibly wealthy might be extremely charitable and provide food, clothes, shelter to the poor…

            No no no, they’re not monsters! They’re practically saints!

            Burt wanted to go outside and take a walk. Weather permitting, we tried to walk every evening. We left the cafeteria and departed through the main door along with a stream of other people.

            The building we exited was another one of the terrafoam projects. Terrafoam was a super-low-cost building material, and all of the welfare dorms were made out of it. They took a clay-like mud, aerated it into a thick foam, formed it into large panels and fired it like a brick with a mobile furnace. It was cheap and it allowed them to erect large buildings quickly. The robots had put up the building next to ours in a week.

            The government had finally figured out that giving choices to people on welfare was not such a great idea, and it was also expensive. Instead of giving people a welfare check, they started putting welfare recipients directly into government housing and serving them meals in a cafeteria. If the government could drive the cost of that housing and food down, it minimized the amount of money they had to spend per welfare recipient.

            Downstairs there was the cafeteria staffed by robots. The robots were not bad — the food was acceptable. They also kept the bathrooms, hallways and rooms spotless. Every day at 7AM, 12 PM and 6 PM the breakfast, lunch and dinner meal shifts began. There were six 15-minute shifts per meal to save on cafeteria space. Burt and I had the third shift. You sat down, food was served, you ate, you talked for 5 minutes while you drank your “coffee” and you left so the next shift could come in. With 24,000 people coming in per shift, there was no time for standing in a cafeteria-style line. Everyone had an assigned seat, and an army of robots served you right at your table.

            Each terrafoam dorm building had a four-acre foot print. It was a perfect 417 foot by 417 foot by 417 foot solid brown cube. Each cube originally held exactly 76,800 people. Doubling this to 153,600 people in each building was unthinkable, but they were doing it anyway. On the other hand, you had to marvel at the efficiency. At that density, they could house every welfare recipient in the entire country in less than 1,500 of these buildings. By spacing the buildings 100 feet apart, they could house 200,000,000 people in a space of less than 20 square miles if they had wanted to. At that density, they could put everyone in the country without a job into a space less than five miles square in size, put a fence around it and forget about us.

            “In accordance with ordinance 605.12b, you have been assigned room 140352 in building 16, resident quant C. This assignment provides you with suitable housing and nourishment to sustain your life. Please board the bus.”

            That was how you ended up in Terrafoam. The system knew you had no means of support, so it “gave” you one. You could leave terrafoam once you regained a means of support, but there really was no way to do that unless Manna gave it to you.

            Was it prison? Yes. But there were no walls. The food was good. The robots were as nice and respectful as they could be. You could walk outside wherever and whenever you wanted to. But there was an invisible edge. When you walked too far away from your building and approached that edge, two robots would approach you. I had tried it many times.

            “Time to turn around Jacob Lewis105. There is construction in the next zone and, for your safety, we cannot allow you to proceed.” There were a hundred reasons the robots gave for making you turn around. Construction, blasting, contamination, flash flooding, train derailments, possible thunder storms, animal migrations and so on. They could be quite creative in their reasons. It was all part of their politeness. If you turned around you were fine. If you made any move in any direction other than the one suggested, you were immediately injected and woke up back in your room. I had only tried it twice.

          • Mark Bahner says:

            Burt wanted to go outside and take a walk. Weather permitting, we tried to walk every evening. We left the cafeteria and departed through the main door along with a stream of other people.

            That whole passage just seems so implausible to me…with people eating in large cafeterias and living in dormitories.

            Scenarios like Terminator (without the time travel, and without the Terminators trying to look human) and The Day After (the TV special way back about the aftermath of a nuclear ballistic missile exchange) are believable to me. People being forced to eat in cafeterias and live in dormitories don’t seem believable to me.

          • Nornagest says:

            They weren’t forced, but dormitory-style living was common for lower-class and single middle-class people living in cities before WWII — in styles ranging from flophouses where dozens of people shared a room, through semi-private “cage hotels”, up to residential hotels with private rooms. Meals were sometimes provided. A useful search term for these is “single room occupancy”; zoning standards killed them in most places by the Fifties or Sixties, although a few survived into the late 20th century and even today in places like the Bowery in NY and the Tenderloin in SF. Ever wondered why you’ve never seen anything like Judy Hopps’ apartment from Zootopia in real life? That’s why.

            Cafeteria-style dining is a different story — that only got started around 1900, but appeared in a variety of formats, for most social classes, until about 1960 when fast food displaced it outside of institutional niches. One version that might be salient here is the “automat”, a partially automated cafeteria where cooks in a backroom kitchen cooked meals that customers bought through coin-operated kiosks, without human interaction.

          • Mark Bahner says:

            They weren’t forced, but dormitory-style living was common for lower-class and single middle-class people living in cities before WWII…

            Yes, that’s what doesn’t ring true to me. Supposedly, robots are doing all these wonderful things, but somehow some large group of people is worse off than most people are now. Why aren’t the robots building single-family homes? And why don’t all homes have Rosie the Robot (who is several generations behind the Renée the Robot, but still provides reasonable service?

            I understand that the reason is that there wouldn’t be a story in the scenario I suggest. So I think they should have thrown a previous nuclear war into the story. That would allow the great lines from The Book of Eli:

            People had more than they needed. We had no idea what was precious and what wasn’t. We threw away things people kill each other for now.

            And it would allow the dormitories and cafeterias. 🙂

          • Nornagest says:

            some large group of people is worse off than most people are now

            I’m not sure this follows. SROs are dead, but not because people got better off in this regard — housing expenses as percentage of income went up in the second half of the 20th century. The people that would have lived in them are now living with roommates or in some cases under bridges, which I’m not sure is better.

          • Murphy says:

            @Mark Bahner

            Why do poor people now often live in tenements with mold problems?

            our society is pretty rich, improving the quality of the dwellings probably wouldn’t even be terribly expensive.

            but the people who want that improvement have very little power and the people who’s taxes would go up by 0.001% to pay for it are much more powerful and don’t see it as their duty to make that so. They’ve already provided a roof and food, why should they provide more?

            So, why would the taxpayers of the future be any nicer to you when you’re the one living in social housing?

            The poor people in manna already have a substantially better standard of living across many metrics than some people on welfare today. And as such many would declare it “mission accomplished” and perfectly acceptable.

            “Why aren’t the robots building single-family homes”

            Land might still be valuable to rich people and 1 rich person who wants a slightly larger hunting reserve has more economic power than 1,000,000 penniless people who want a bigger home.

        • Tracy W says:

          what is the capitalist economy based on ?

          Things that are scarce. Holiday spots. Seats at live stage performances. Inherently rare items like original paintings or historic collectibles. Etc.

        • vV_Vv says:

          Why not ? If robots are able to out-perform humans at pretty much every task, including finance; then what is the capitalist economy based on ?

          Capital, as the name says. Now it’s actually capital+labor, once you have robot workers better than humans at everything it will be only capital.

          It can’t be money, because money is essentially worthless now that the robots run the entire financial industry. It could be some sort of rare natural resource, like gold or uranium, but what would you buy with your uranium coins, and who would want them ?

          As long as there is trade which is not barter there is going to be money. Whether it is fiat money, gold coins, uranium coins, bitcoins, or something else, it is a technical question, but you will need some transferable token to represent the opportunity costs of producing and delivering different goods and services.

        • DocKaon says:

          In economics there are 3 factors in production labor, capital, and land. Land in this context means natural resources. Automation replaces labor with capital. Now, you have capital and land with land being the scarce resources. With labor no longer having a role in production, the capitalist economy of making money by selling things to workers disappears. You’re left with a manorial economy where those with land produce what they need using automated systems, including more automated systems.

          Basically, the people who had money before the transition use it to buy control of natural resources and a stock of robot serfs and become a new landed aristocracy. What happens to everyone else depends on the charity of the landowners.

          • Jiro says:

            If by “capitalist” you mean “someone who owns enough capital that he can live a comfortable life off his capital, without selling his labor”, then as robots get cheaper and cheaper, poorer and poorer people can afford robots, which means that the capitalist class has just expanded to include those poor people.

            If a robot costs a dollar and does anything you could ever need, then everyone with a dollar is now a capitalist. Who cares if he can’t sell his labor any more? He has enough capital now that he doesn’t need to.

          • vV_Vv says:

            If a robot costs a dollar and does anything you could ever need, then everyone with a dollar is now a capitalist. Who cares if he can’t sell his labor any more? He has enough capital now that he doesn’t need to.

            The number of robots and the natural resources they can process is still going to be limited, so the guy with one $1 robot is going to compete with the guy with one billion $1 robots.

          • Mark Bahner says:

            The number of robots and the natural resources they can process is still going to be limited, so the guy with one $1 robot is going to compete with the guy with one billion $1 robots.

            “Compete,” in what sense? If the person with the $1 robot can get everything he/she needs from the $1 robot, what is the need to compete?

          • Aapje says:

            @vV_Vv

            Also, the land may be owned by a person. Having a robot that can mine gold has little use if all the viable gold veins are owned by others.

        • Deiseach says:

          If robots are able to out-perform humans at pretty much every task, including finance; then what is the capitalist economy based on ?

          That’s a bit like saying “if slaves can replace paid workers, then what will the economy of the slave-owners be built on?” Yes, the Southern states did eventually have a dead-end economy but on the other hand, people who wanted to make the most returns on their investments plainly saw some benefit in starting off a slave trade in the first place, rather than relying on hiring workers or trying to import free Africans to work for them, and fortunes were being made from using slave labour.

      • One very important reason our welfare system isn’t more generous is that we, as a society, need people to work and they often won’t work if we provide them with a good alternative. If robots are doing all the work that’s a good reason to do a lot more government redistribution.

        • vV_Vv says:

          If robots are doing all the work that’s a good reason to do a lot more government redistribution.

          And what stops the robot owners from seizing the government and abolishing all welfare?

          Capital owners can’t do it already because: 1) people who are on welfare have family and friends who work, and can therefore strike, or boycott, or otherwise disrupt the operations of the capitalists, and 2) because people can turn to violence.

          But once the capitalists don’t need workers, or customers who aren’t themselves capitalists and once they have unbeatable private robot police and military, what incentive do they have to care for the plebs?

          • Chris Williams says:

            This is a short, fictional story, that aims to explore exactly the topic you’ve mentioned. It talks about both the approach that robots could put us on vacation forever, or that they could be used by the rich to cast aside most of society. It isn’t much longer than this article and definitely worth the read: http://marshallbrain.com/manna1.htm

          • Mark Bahner says:

            “And what stops the robot owners from seizing the government and abolishing all welfare?”

            Basic humanity. I didn’t follow the Hunger Games movies closely, or read any of the books, but the thing that struck me is that the people in The Capitol were caricatures of humans. What human being would watch children suffering and dying and not want to help?

            Look at the list of the Forbes 400. Can you imagine them “seizing the government and ending all welfare”? I sure can’t!

          • The Nybbler says:

            Look at the list of the Forbes 400. Can you imagine them “seizing the government and ending all welfare”? I sure can’t!

            #2, #4, #5, #6 (both), #15, #20, #27, #35, the first #39, the first #91, #115, the first #167, #226, #248 (both), and of course #315.

            Most of the rest I’m not familiar with.

          • Mark Bahner says:

            #2, #4, #5, #6 (both), #15, #20, #27, #35, the first #39, the first #91, #115, the first #167, #226, #248 (both), and of course #315.

            Without a scorecard, I can’t tell the players. 😉 Is this the Forbes 400 we should be working with:

            https://www.forbes.com/forbes-400/#6bf32d4e7e2f

            ?

            If so…let’s just start at the beginning of your list: Jeff Bezos (#2)? Mark Zuckerberg (#4)?

            I don’t know the details, but I think Mark Zuckerberg is attempting to set up a charity to rival the Bill and Melinda Gates Foundation. Look, he and his wife are spending $3 billion to “cure all disease”!

            http://www.businessinsider.com/mark-zuckerberg-cure-all-disease-explained-2016-9

            Putting the technical possibility of the project aside…how nice is that?! Isn’t that equal to or close to the most wonderful thing anyone could even conceive of doing?

            And Jeff Bezos…who’s spent millions supporting gay marriage and most money supporting Democrats? Why would a person who wants to “abolish all welfare” support mostly Democrats?

          • AG says:

            As per Chris Williams’ contribution, Cory Doctorow’s novel “Walkaway” also explores that period of transition. The thesis is that people need to choose to create the post-scarcity society, it is far from inevitable based on the technology alone.
            (Akin to arguing that a successful democracy like America was not inevitable. If the American Revolution had failed, there would not have been an equivalent nation to create a similar world history. The specific factors involved in the found of the US, including key leaders guiding its development in the early years, was critical.)

            Or, “we must choose to create and maintain a culture of cooperating, where the absence of intervention will be a culture of defecting.” And that applies to the usage of hypothetical scarcity-ending technology.

          • vV_Vv says:

            Basic humanity.

            How much did you spend in entertainment last year? Books, movies, videogames, etc. How about fancy food? Pleasure travel? Other first-world luxuries?

            How many stereotypical starving African children could you have feed with that money?

            In the robocapitalist future you’ll be starving while the fertile land next to you is being robofarmed to make environment-friendly rocket bio-fuel for the elite’s space yachts.

            Sounds unlikely? Well, in San Francisco there is a robot owned by a animal shelter, of all things, that patrols the streets and harasses homeless people. Seriously. Make that thing a Terminator and multiply it by one billion and you’ll get the idea of how it’s going to be.

            I didn’t follow the Hunger Games movies closely, or read any of the books, but the thing that struck me is that the people in The Capitol were caricatures of humans. What human being would watch children suffering and dying and not want to help?

            The Romans for instance, who were the overt inspiration for the civilization of the Hunger Games.

            Who knows, maybe after complete labor automation, being a gladiator who engages in deadly combat could be one of the few jobs that a human could still do.
            But even if morality doesn’t shift that far, normal indifference could be enough to cause misery and death to anybody who isn’t an aristocrat.

          • Thegnskald says:

            I’m not a huge fan of class, but rich people are still human.

            And assuming resources are as plentiful for them as we are saying, it would take one rich person to keep living standards decent for everybody, even assuming the worst kind of economic apocalypse.

          • Mark Bahner says:

            How many stereotypical starving African children could you have feed with that money?

            Yes, but those starving African children aren’t my neighbors. And I don’t have even millions of dollars, let alone billions or trillions.

            Again, I point to Bill and Melinda Gates. They do have billions of dollars, and they’re giving billions of dollars to help the starving/sick children of Africa.

          • INH5 says:

            One person getting handed enough wealth to buy a lavish lifestyle for themselves and either provide all of their fellow countrymen with a decent lifestyle or hire a huge mercenary army to take whatever they want from their fellow countrymen is pretty much exactly what happened in places like Saudi Arabia, Qatar, and Brunei. The rulers of those countries didn’t even need to seize control of the government, because they were already absolute monarchs. From the data that we currently have, providing your citizens with a decent lifestyle through infrastructure spending and a generous welfare state seems to be a much more popular choice than “hire a huge mercenary army.”

            Not that those countries don’t have problems – the Saudi religious police would not be possible without the oil wealth, either. But “the vast majority of citizens starving while the king hordes all of the oil money for himself” is clearly not one of them.

          • Mark Bahner says:

            The Romans for instance, who were the overt inspiration for the civilization of the Hunger Games.

            Yes, and how many years has it been since the Romans watched gladiators in the Colosseum? And how rich were the people in the crowd watching the action?

          • Mark Bahner says:

            Sounds unlikely? Well, in San Francisco there is a robot owned by a animal shelter, of all things, that patrols the streets and harasses homeless people. Seriously.

            Here’s what the SPCA described:

            SF SPCA staff members said the facility had been plagued with break-ins, staff members had been harassed as they went to the parking lot and sidewalks were littered with hypodermic needles. Jennifer Scarlett, the SF SPCA president, said in a release that her organization “was exploring the use of a robot to prevent additional burglaries at our facility and to deter other crimes that frequently occur on our campus – like car break-ins, harassment, vandalism, and graffiti – not to disrupt homeless people”.

            How is that similar to billionaires or trillionaires keeping money to themselves while people around them starve?

            In fact, how can you even describe the situation as being that the SPCA is harassing homeless people? The SPCA basically has a parking lot and a building. They aren’t sending a robot to chase homeless people down the street.

          • vV_Vv says:

            @INH5

            One person getting handed enough wealth to buy a lavish lifestyle for themselves and either provide all of their fellow countrymen with a decent lifestyle or hire a huge mercenary army to take whatever they want from their fellow countrymen is pretty much exactly what happened in places like Saudi Arabia, Qatar, and Brunei.

            But the Gulf monarchs still need a population of subjects to support their lifestyle and not behead them. Once they’ll get cheap unbeatable robots that always do what they are told and never join al-Qeda or switch sides to one of their hundred cousins, what will they need subjects for?

            @Mark Bahner

            Yes, but those starving African children aren’t my neighbors.

            While Zuck and Bezos are?

            I mean, you don’t even have to look at the ultra-rich, in the US even the upper-middle class often segregates itself into gated communities and college campuses (which are basically charter cities) to stay away from the plebs.

            Again, I point to Bill and Melinda Gates. They do have billions of dollars, and they’re giving billions of dollars to help the starving/sick children of Africa.

            They live in a society where it is high status to do so. What will be the value system of the elite in 100 years? in 1000 years? Will a civilization where most people depend on the benevolence of a few elite be sustainable?

            In fact, how can you even describe the situation as being that the SPCA is harassing homeless people? The SPCA basically has a parking lot and a building.

            So land is being allocated by society to provide facilities for a first-world luxury supported by upper-middle class people, and economically unproductive people are chased away from it, to be poor and miserable somewhere else where these upper-middle class people can’t see them. That’s the point.

          • INH5 says:

            But the Gulf monarchs still need a population of subjects to support their lifestyle and not behead them.

            No they don’t. The vast majority of the work that needs to be done in those countries is already done by migrant workers. About 88% of Qatar’s population consists of expatriates, for example. Most of the citizens that have jobs have government jobs that pay more than any migrant worker could hope to make, which is to say that they’re effectively make-work jobs.

            As for the risk of being beheaded, considering how much terrorist financing comes from the wallets of private Gulf State citizens, the redistribution of oil wealth clearly contributes to that problem even while it removes some of the incentives to rebel. Osama Bin Laden himself only had the resources to found Al-Qaeda because his father had become a billionaire by securing exclusive construction contracts with the Saudi royal family. Considering this track record, it seems likely that large mercenary armies would have bought more security for less money than the welfare states have.

          • vV_Vv says:

            No they don’t. The vast majority of the work that needs to be done in those countries is already done by migrant workers. About 88% of Qatar’s population consists of expatriates, for example.

            Yes, but the key jobs (security, but also managing large companies) are in the hands of nationals, and can’t be effectively given to immigrants without compromising sovereignty.

            Most of the citizens that have jobs have government jobs that pay more than any migrant worker could hope to make, which is to say that they’re effectively make-work jobs.

            Some of these people are relatives of those who do actual work (remember that the Arab concept of “relative” is much broader than the Western one), others, probably most, are given these make-work jobs or other welfare just so they don’t revolt.

            As for the risk of being beheaded, considering how much terrorist financing comes from the wallets of private Gulf State citizens, the redistribution of oil wealth clearly contributes to that problem even while it removes some of the incentives to rebel. Osama Bin Laden himself only had the resources to found Al-Qaeda because his father had become a billionaire by securing exclusive construction contracts with the Saudi royal family.

            The Gulf States are barely politically stable. They haven’t collapsed yet because of the large redistribution of oil wealth, but even with it there is unrest which manifests itself as support for Islamic terrorism. Arab countries that don’t have oil wealth to redistribute, such as Yemen, Lebanon and Palestine are in fact in near-constant civil war.

          • anonymousskimmer says:

            But the Gulf monarchs still need a population of subjects to support their lifestyle and not behead them.

            Some of these people are relatives of those who do actual work

            You answer your own statement. One doesn’t go around beheading members of one’s own house, just as one doesn’t go around rounding up and dorming one’s cousins. At least not in Arabia. These people still have a clannish religion that they believe in, after all.

            Foreigners are another thing entirely.

      • JulieK says:

        Has there ever been a human society whose products flowed “evenly to everyone?” A better question is whether everyone gets at least some part of the lack of scarcity.

    • albatross11 says:

      Whether it’s an AI or a capitalist turning me into paperclips because I don’t meet any other of his needs and he has the power to do so, I’m pretty-much equally unhappy with the outcome.

    • Kyre says:

      The arguments in this thread have been going through my head for the last 10 years or so … and yet I’ve never seen them written down anywhere.

      Once the value of human labor goes below subsistence level, one hopes that either democracy or charity prevents starvation of people who don’t own enough capital to support themselves.

  3. Bugmaster says:

    We’ll just get people being pushed into worse and worse jobs, in a way that does not inspire widespread sympathy or collective action.

    I don’t know much about economics, so the following is just conjecture, but: I think one reason for this state of affairs might be the reduced pace of technological innovation. In the past, when an industry (such as horsemanship, agriculture, switchboard operation, etc.) was automated, the technological advancements that led to this automation also led to the creation of several new industries. To use an extreme example, the same technology that eliminated switchboard operators led to the massive paradigm shift that was the Information Age.

    However, in the modern Western world, new industries are created a lot less frequently; and thus workers whose jobs got automated have nowhere to go but down. This could be due to a combination of factors:

    1). Automation today is increasingly the result of incremental advances in the field, as opposed to paradigm-shifting discoveries. Thus, most of the new industries that could’ve been created, already have been.
    2). Technological progress in general has slowed down significantly, because…
    3). …All the easy stuff has already been discovered, and/or…
    4). …A combination of the socio-political climate, as well as extremely aggressive copyright/patent law, is causing a chilling effect (at least, in the Western world).
    5). Western countries are suffering from resource depletion; it’s harder to innovate on zero budget.

    • Calion says:

      I think you’d have to back up this “new industries are being created less frequently.” What counts as an “industry”? If “software” is one industry, then sure, maybe. But that’s like calling “manufacturing” one industry. And doesn’t ride-sharing count as an industry, to name just one of many examples?

      • Yosarian2 says:

        Is it? Or are Uber/Lyft the same industry as taxis, just using automation to replace the taxi dispatcher jobs with an app?

        • Mark Bahner says:

          Or are Uber/Lyft the same industry as taxis, just using automation to replace the taxi dispatcher jobs with an app?

          In less than a decade, when Uber/Lyft cars don’t have any drivers in them, I think that will be a new industry. 🙂

          • anonymousskimmer says:

            Rent-a-horse.

            Or a sidewalk replacement (why build sidewalks when everyone can just hail an Uber to get from store to store?).

          • Mark Bahner says:

            Or a sidewalk replacement (why build sidewalks when everyone can just hail an Uber to get from store to store?).

            Well, that’s another thing. With computer-driven vehicles, there will be no brick-and-mortar stores. People will order via smartphone, and goods will be delivered with computer-driven delivery vehicles.

            That will replace not only the sidewalks at the stores, but also the parking lots. (Not to mention cashiers and everyone else in a traditional store.)

      • Deiseach says:

        And doesn’t ride-sharing count as an industry, to name just one of many examples?

        No, it’s somewhere in between its roots in “if you have to drive into the city for work anyway, make a few dollars on the side giving people a lift” and traditional service industry jobs. I think most people would class Uber drivers as now pretty much employees of Uber, not freelancers using Uber to hook them up with potential customers, and from what I read of people’s interactions with drivers when using Uber they don’t seem to treat them as “This is Jose or Maria who is using their own car to give me a lift”, they treat them as “the Uber driver”, so Uber-style work is functioning as a hackney service, whatever Uber may do with its “the drivers are all independent contractors”.

        And Uber is plenty interested in self-driving cars to cut out that much of a middleman in providing a service to customers – can it really be called “ride-sharing” any more when there’s no human driver and not even the pretence that you’re sharing a ride into town with the guy who was going there anyway?

        • Mark Bahner says:

          …can it really be called “ride-sharing” any more when there’s no human driver and not even the pretence that you’re sharing a ride into town with the guy who was going there anyway?

          I think transportation-as-a-service probably will lead to more ride-sharing, at least up to the time people are infinitely wealthy…or killed off by robots. The reason I think transportation-as-a-service will lead to more ride-sharing is that it will be less expensive per mile to ride-share.

    • BlindKungFuMaster says:

      I know Peter Thiel likes to go on about the decrease in innovation, I would love to see some solid numbers on that.

    • keaswaran says:

      Alternatively, you might think the new industries are being created just as quickly, but there’s a lag of 15 years for industry creation as you have to train a whole new generation of kids to participate in these industries. When industry destruction comes once every few decades, a 15 year lag is fine. But when industry destruction comes faster and faster, the fixed cost of 15 years becomes more significant.

    • seladore says:

      In the past, when an industry (such as horsemanship, agriculture, switchboard operation, etc.) was automated, the technological advancements that led to this automation also led to the creation of several new industries […] However, in the modern Western world, new industries are created a lot less frequently

      Surely this is an illusion based on the fact that new industries take time to fully develop? What might have seemed like a small development 100 years ago has grown into an entire field over the course of a century. So of course the developments we see around us don’t seem as significant.

      It’s like looking at a tree and concluding that the tree’s growth has slowed down significantly. Because 100 years ago the tree was growing whole new gigantic branches, but all the growth in recent years has been at the twig level.

  4. dansimonicouldbewrong says:

    Trump voters might have a different explanation for the steady fall in wages and employment rates for unskilled workers since the 1960s–something that rhymes with “in big station”…

    • Scott Alexander says:

      See eg http://www.hbs.edu/faculty/Publication%20Files/09-013_15702a45-fbc3-44d7-be52-477123ee58d0.pdf , “Within the large empirical literature looking at the effects of immigration on native employment and wages, most studies find only minor displacement effects even after very large immigrant flows.”

      • azhdahak says:

        I really doubt this.

        I work in an expensive city with a lot of immigrants. The company has the same operation I work at set up in a cheap city with few immigrants. I’ve talked to people who worked out there and moved, and they make about 1.5x as much as we do.

        • Brad says:

          You are confident that your personal anecdotes can be blindly extrapolated to give a better picture of reality than a large empirical literature?

          • vaniver says:

            Note that the literature here is often deeply flawed–for example, for almost 30 years Miama during the cocaine era has been used to show “look, a bunch of immigrants at once is fine,” [Card 1990] but as soon as someone asked the question “but what about for the people the immigrants are competing with?” the answer became “oops, I guess maybe not.” [Borjas 2017]

            This is, of course, contested; see this article by Vox: https://www.vox.com/the-big-idea/2017/6/23/15855342/immigrants-wages-trump-economics-mariel-boatlift-hispanic-cuban , which claims that the decline is because the trend was negative before the boatlift then flattened afterwards. (Thankfully, it doesn’t make the sample size criticism, which can be immediately turned on the original Card study as well.)

            If it were anecdote that flew in the face of theory, that would be one thing; but the issue is that anecdote and theory are on the same side, with the empirical literature on the other. It seems wise to suspect the literature in such cases.

      • j r says:

        There are two reasons why immigrants haven’t had a big impact on native employment and wages. The first reason is that the reserve wage for native workers is pretty high. Almost no American is going to work for less than minimum wage. Finding good workers at the minimum wage is hard enough. Native-born Americans aren’t lining up to wash dishes in restaurants or even to be laborers on construction sites, which pays relatively well. In the absence of workers willing to do low-wage work, employers don’t just raise wages until they find workers; they mostly invest more money in capital to automate those functions or those functions simply go undone. The second reason is that immigration follows a pattern. Immigrants go to where unemployment is low (i.e. they go to tight labor markets). Immigrants go to where the competition from native-born workers is the least. This makes perfect sense.

        The narrative of immigrants replacing native workers is mostly a fairy tale. There are some sketchy stories about replacing native born workers in relatively less-skilled tech areas with workers on H1B visas, but that’s such a small number that it can’t explain the trend.

        • Jiro says:

          In the absence of workers willing to do low-wage work, employers don’t just raise wages until they find workers; they mostly invest more money in capital to automate those functions or those functions simply go undone.

          This amounts to “if employers can’t get cheap immigrants, they get cheap machines instead”. This means that you really should think of technological unemployment and immigration as basically the same thing causing the same problem. You’ve correctly deduced that opposing just one of them is useless, but not that opposing both of them is.

          If I have nails in my driveway and nails along the road, I shouldn’t say “it’s pointless removing the nails along the driveway because the ones on the road will still puncture my tires, and it’s pointless removing the nails on the road because the ones in the driveway will still puncture my tires”.

        • Ratte says:

          The narrative of immigrants replacing native workers is mostly a fairy tale.

          And yet…

          Short version is that once the crackdown on illegal labor hit a major bakery wages for legal workers rose a staggering 40%.

          • The Nybbler says:

            Short version is that once the crackdown on illegal labor hit a major bakery wages for legal workers rose a staggering 40%.

            That’s based on

            According to a former consultant to the bakery, MSI paid the black workers $14 an hour, versus the $10 an hour the Mexican workers were making through Labor Network.

            But MSI itself says otherwise:

            Ed French, owner of Elgin-based Metro Staff Inc., says his company became the main provider of workers for the bakery and that about 80 percent of them are black. According to French, workers at the bakery were paid slightly less before his company was hired two and a half years ago — with wages up by about 25 cents an hour, to just above minimum wage.

          • anonymousskimmer says:

            @The Nybbler

            Staffing companies make their profit by taking a percentage of what the employing company pays for each worker. It looks like Labor Network was taking a much larger percentage than MSI is, possibly because they knew most of their employees were illegals, and thus not in a position to bargain?

            Or, re-reading, the employees always got paid about minimum wage, and the $14/$10 amount is what the bakery paid the staffing companies. This is more likely, as the bakery consultant is in no position to know what the staffing companies are actually paying the workers, they would just know what the bakery is paying the staffing company per worker.

            Which makes me wonder what else Labor Network was skimping on. I could easily see this amount as the difference between part-time labor and full-time labor which is guaranteed ACA-compliant health care.

      • Tenacious D says:

        What about the impact of undocumented workers on the quality of data used to address questions like the one in this post? I’m assuming, perhaps naively, that the graphs above probably are not all consistent in the way they include/exclude them which could contribute +/- 5 million prime age males if common estimates are correct.

      • Thyle Dysig says:

        On average it may not have a huge effect, but it may have an effect much like the one you have carefully laid out. Some jobs will be effected, but others will not. Living in a state with a lot of immigrants, I can tell that they have come to dominate some jobs, such as construction. We had a twenty year period where the wages for construction workers did not change. This pushed a lot of construction workers into other jobs.

        Given that economics is based on supply and demand, increasing the supply of workers has to have an effect. And unless the workers evenly distribute throughout the economy, the new supply is going to change the dynamics of several industries. This might hurt, or it might help, but it will have an effect.

      • Edward Scizorhands says:

        “Within the large empirical literature looking at the effects of immigration on native employment and wages, most studies find only minor displacement effects even after very large immigrant flows.”

        I was going to ask “are those ‘minor displacement effects’ over the whole labor force, or concentrated.”

        And I look up your quote, and the very next sentence is

        On the other hand, some more recent studies have found larger effects, and many studies note that the negative effects are concentrated on certain parts of the native population.”

        And the next sentence after that describes what those certain parts are, and they are exactly what you think they are.

      • AG says:

        Counter-link:

        Shortage Of Illegal Labor Caused Construction Worker Wages To Rise Up To 30%

        I remember a similar case in Agriculture, too. So of course some of the people there are using DACA to talk about the importance of migrant workers now, while American seasonal workers aren’t getting paid enough to live on even if they work year-round by constantly moving to the current season.

    • I did some research on this too and like many things pertaining to economics (such as the debate over unemployment vs. minimum wages, etc.), the answer is inconclusive. That’s the problem with economics…so little can be answered definitively. Some studies say it’s bad, others no, others good, etc. So we don’t know. High-IQ immigrants create jobs, such as seen in Silicon Valley. Immigrants consume, which helps businesses. Sometimes immigrants ‘take’ jobs that would otherwise go to a native worker, but other times labor is not interchangeable, meaning that an immigrants who can do a job for $10/hour is not identical as a native-born citizen that can do it for $12, save for the $2 price difference. There are differences in quality and other factors.

    • dansimonicouldbewrong says:

      From the same document you linked to: “While large, economy-wide displacement effects appear unlikely, it is still possible that specific sectors or population groups experience significant impacts from immigration. Studies evaluating the potential displacement effects for the at-risk groups or sectors, especially those with strong empirical identification, would still have a place in the vast displacement literature.”

    • anonymousskimmer says:

      “in big station”

      This made me think of stagflation. It wasn’t until Scott’s reply that I realized what you were talking about.

  5. Calion says:

    You commit an error here that i’ve seen before:

    “In 1790, 90% of Americans were farmers, even though agricultural productivity had been improving for ages. Today, 2.6% of Americans are. Maybe manufacturing just had the same kind of moment. Advances in technology can put farmers out of work (but shift them to manufacturing).”

    The problem here is that 2.6% of adult Americans today (about 6 million people) is more than 90% of adult Americans in 1790 (less than 3 million people). So it’s not necessarily the case that anyone actually had to lose their job in this transition. Maybe they did; I don’t know (I mean, we all know the stories of family farms going bust in the last thirty or forty years, so obviously it did happen *some*). But there’s no reason to assume so.

    • Simurgh says:

      You are of course correct to point out that the absolute numbers did not go down, but I think this is not relevant for the main argument. Increases in agricultural productivity enabled a massive population growth, but this is not likely to be repeated in other domains.

    • Mark Bahner says:

      The problem here is that 2.6% of adult Americans today (about 6 million people) is more than 90% of adult Americans in 1790 (less than 3 million people). So it’s not necessarily the case that anyone actually had to lose their job in this transition.

      Per the NYT, the farm population in 1920 was 32 million:

      U.S. farm population through history

  6. algorizmi says:

    Suppose some jobs are impacted in the other direction à la Jevons paradox. Credentialism implies a long lead time to retrain new hireable employees. If there is not already a larger pool of skilled laborers than jobs this implies wage volatility.

    What jobs are wages increasing fastest for? A good place to look for positive employment effects of technology.

  7. meh says:

    Graphs comparing college to high school over time need to show percent of the population each group is over time. (i vaguely remember this coming up on the blog before)

    • Cliff says:

      Yes, I was very surprised Scott did not mention this as it has a huge effect on exactly what he is discussing in that part. Of course as college attendance goes up, the high school only crowd becomes more and more the dregs of society, with higher unemployment and pretty much every other negative metric, without any underlying change

  8. Hamish says:

    The self-delusion theory (for selecting “disabled” vs “looking for work” in surveys) seems implausible, what about an alternative explanation: Diminishing marginal value of employees vs robots lead employers to skimp on safety, causing accidents that resulted in disability in employees.
    Another possibility is that they literally intentionally made themselves disabled by themselves skimping on safety to be entitled to disability payments

    • Surely the number of crippling workplace accidents has not increased over the past century?

      • Hamish says:

        It could be a thing along the lines of military casualties with way fewer deaths, and way more people coming back injured. Modern medicine seems to be good at keeping people alive, but not keeping them in a good state necessarily,

        • Tenacious D says:

          It’s not just modern medicine. Safety practices have also come a long way. For example, proper lock-out procedures for maintenance tasks have surely prevented a large number of disabling injuries.

        • Ratte says:

          I know that (at least in the military) the definition of ‘disability’ has vastly expanded; getting diagnosed with sleep apnea, for example, can mean a check for 50% of your pay for life. Not surprisingly, this has caused a spike in diagnoses.

    • Aapje says:

      @Hamish

      What about an alternative alternative explanation: demands on employees have been going up, so more and more people are ‘disabled’ in the sense that they cannot meet those demands.

      Ultimately, ‘disabled’ is not an objective term. I’m disabled for the job of professional speed skater. I can never measure up no matter how much effort I put in.

      Perhaps the influx of women allowed employers to raise their demands, replacing the least capable men with more capable women.

      • Chris Williams says:

        Part of this is also the way the social safety net is setup. I know a dude I went to high school with years ago. He’s bipolar and a schizophrenic (sp?). Unmedicated, he basically goes all Charlie Sheen crazy from time to time. It makes him unable to function in society.So he’s considered disabled. However, he responds very well to medication. Now that he’s declared disabled and gets medication and doctor visits/therapy for free through Medicaid, and is on meds, he’s turned into a completely normal guy. 95% normal anyway. It’s completely turned his life around. Now, he would be perfectly able to hold down a job, or go to school, or learn a trade. That’s the rub though. If he gets a job or goes to school it serves as evidence he’s not disabled and could get him kicked off of food stamps, public housing and medicaid. So it’s a safer deal for him to stay home and play xbox than to contribute to society. He’d rather be “normal” in the head and sitting around than being crazy – which are basically his choices. If our social safety net was better designed, we could get some of those people back in the workforce.

        • Brad says:

          There are transitional provisions which are designed to serve as a bridge in exactly this type of situation which would allow him to remain on assistance for a period of time until he was securely in a job which would hopefully include healthcare coverage or at least a high enough pay when combined with ACA subsidies to get healthcare that way.

          There’s still the issue of the very high implicit marginal rate which means that a job has to be fairly well paying to “worth it” but the kind of instant catch-22 you are talking about is something that people have thought about and tried to prevent.

  9. Calion says:

    Finally made it to the end. I feel like you’re not asking the really big questions here. What actually happens when robots can literally do any job? So they do those jobs, and nobody has work, and so…nobody can buy anything? And so the rich people who own these robots don’t make any money, and everybody starves and dies among unimaginable wealth? That seems…unlikely. So what will happen? Once we have the answer to that, it seems like what happens with less than total robotization will be somewhere between here and there. I have some undeveloped ideas on this, but I am disappointed that the question wasn’t even really raised here.

    • Scott Alexander says:

      By the time robots can do literally any job, robots are going to be the ones in charge of deciding what happens next.

      I agree this is an important issue, but I have a bunch of other blog posts about it. Today’s is about the short-term.

      • Jon says:

        Try economist Don Boudreaux on the subject. He’s very optimistic about our future of unimaginable wealth.

        https://www.google.com/search?q=site%3Acafehayek.com+robots

        • sharper13 says:

          Here’s a real life example related to the whole robots can do everything.
          I’ll give you my perspective, but here’s an interesting story about it from a 1985 perspective, which is right about the tipping point.

          Let’s say a significant number of people work in clerical tasks. These include storing and retrieving information, adding, totaling and reporting sets of numbers, etc… They’re mostly employed by businesses large and small. Virtually no one except the super-rich employ them privately, everyone else at most rents their services as part of a business. Who could afford their own clerk to live at their house, when they can’t even afford a nanny.

          Then let’s say a robot+AI comes along which does all of these clerical tasks so much better that no human can compete with it. To start with, businesses snap them up, as they have obvious uses for them. Very few individuals own them, mostly as curiosities for the wealthier. Sure, there’s jobs and work in creating these robot+AI devices, but millions of clerical workers have to find something else to do in order to earn money.

          Now, what are the odds in this situation that the very rich and businesses end up owning all of these robot+AI devices and the former clerks don’t own any?

          Cheating via the use of a time travel device, the risk is 0%. Why? Well, a few reasons.
          1. The wealth gained by the widespread introduction of the personal computer+mass electronic storage means that even our former clerks can afford one as market mechanisms drive their cost down to the marginal cost over time.
          2. Widespread technological advancements in productivity aren’t limited to use by their creators. Once technology reaches a particular level, it’s relatively easy for someone to see what something is doing and make something else to do something similar. (This is why many technologies end up getting simultaneously invented in different places… the base technology/knowledge for that advancement exists.)
          3. Microsoft, Apple. IBM and Amiga didn’t get together and take over the world. They disrupted some markets, but they basically worked within the existing legal and political frameworks. As long as more than one single individual can make robots, they’ll work the same way. Sure, Amazon right now has lots of robots…. but so does the U.S. government, UPS, Walmart, etc… they can’t just take over the world because they are the only ones who own future robots. Everyone will have robots, eventually.

          It’s that last point which is key. Everyone will have robots, eventually. Just like I am typing this on one of the approximately 20 personal computing devices in my five person household, if robots are super-cheap and can replace humans in most tasks, everyone will have a set of personal servant robots to do whatever they want done. Who really cares that much if someone else can afford twice as many, if you have zero now and three of them later? Use one of your three to grow you some food out back in the garden (no signs of super overpopulation coming, so plenty of land out there), another to work on your house (does he have to make the materials himself from dirt and call it adobe? Who cares…) and the third for misc. chores (cheaper than individual washer, dryer, dishwasher, etc…), then do whatever you want with your leisure time.

          I currently have more information storage, retrieval and analysis capacity at my fingertips than entire countries did not that long ago, historically speaking. Has that changed the world? Sure it has, but for the better in so many ways. The more advanced and cheaper robots will do the same.

      • Thyle Dysig says:

        By the time robots can do literally any job, robots are going to be the ones in charge of deciding what happens next.

        Yep. Just like the Angles and Saxons were brought in to do the fighting the Britons didn’t want to do. Eventually, the Angles and Saxon were the ones making the decisions. That’s why we speak a language named after the Angles.

    • fortaleza84 says:

      What happens is that almost everyone collects welfare, whether it’s called “disability,” “basic income guarantee,” “unemployment insurance,” or something else.

      Instructive in the regard is Israel which has a booming economy and (according to Scott’s chart) the highest rate of male non-participation in the labor force. Presumably the reason for this is that a lot of men in Israel receive a stipend and study the Talmud all day long. It’s widely assumed that such a situation is unsustainable, but with improvements in technology it doesn’t need to be.

      • moscanarius says:

        Welfare would eventually have to come by another name in the form of things like research scholarships, money for cultural projects, and useless government jobs.

        Basically, what we already do – just on steroids.

    • vV_Vv says:

      So they do those jobs, and nobody has work, and so…nobody can buy anything?

      Rich people own capital: natural resources, factories and the robot themselves, these things are needed to make stuff that rich people consume, therefore rich people can still trade between themselves even if there is no human labor.

      In the extreme case, there there is one rich person left, or only a family or small group that shares stuff, therefore there is no trade, the robots just extract resources from the environment and process them to serve the needs of their masters.

      The only way for a large population of people who are not capital owners to partake of a labor-less economy is welfare. Yet there seem to be no incentive for private capital owners to accept it: they don’t need workers, and they have a robot army to defend their property and in fact seize all the property from people who can’t afford killer robots.

      • Mark Bahner says:

        The only way for a large population of people who are not capital owners to partake of a labor-less economy is welfare. Yet there seem to be no incentive for private capital owners to accept it: they don’t need workers, and they have a robot army to defend their property…

        Suppose I had the choice between having a $1 billion dollars and needing a robot army to protect myself, and $100 million, and not needing a robot army to protect myself because I’d given the $900 million away. I’d certainly choose the latter.

        Even reducing all those numbers by a factor of 100 or 200, I’d still choose the latter.

        • The Nybbler says:

          Nope, you still need the army even if you give away the 900 million. Now you’ve got a bunch of people who not only have nothing productive to do, but they’re probably reasonably well-fed and healthy… first thing they’re going to do is go after the remaining 100 million.

          • Antistotle says:

            And there’s socialism in a nutshell.

          • Mark Bahner says:

            Nope, you still need the army even if you give away the 900 million. Now you’ve got a bunch of people who not only have nothing productive to do, but they’re probably reasonably well-fed and healthy… first thing they’re going to do is go after the remaining 100 million.

            How many people on the Forbes 400 have an army? They’ve got a lot more than even $1 billion…let alone $100 million.

          • vV_Vv says:

            How many people on the Forbes 400 have an army?

            I’m pretty sure they all live constantly guarded by private armed security, and more indirectly, there is the police force and army of whatever country they live in, which are committed to defend their property rights, and obey to politicians who take their donation money.

            Sure, there is a trade off resulting in taxes, because politicians also have to win those pesky elections, but once you remove the economic incentive to maintain democracy…

          • Mark Bahner says:

            I’m pretty sure they all live constantly guarded by private armed security, and more indirectly, there is the police force and army of whatever country they live in,…

            Yes, and how many mobs do you see in developed countries, storming rich peoples’ mansions for money? Right now, the U.S. per capita GDP is a little over $50,000. In this hypothetical robot future, the U.S. per capita GDP could easily be 4+ times that much (adjusted to 2018 dollars). So it would be easy to give a minimum income of $40,000 to every man, woman, and child in the U.S. What family of 3-4 with an income of $120,000-150,000 is going to storm rich peoples’ mansions for more money?

    • ilikekittycat says:

      Given the history of state power in the 19th/20th centuries, I’d say short-term concessions and social programs, long-term starvation/forced mass migrations/eliminationism. Once you have masses of people that can’t contribute productive labour, they actually look a lot like the class of parasites Hitler painted the Jews as, contributing misery at best and violent revolution at worst. In America we only have to look back to the Okies and Arkies to see how quickly the status of a population can fall. Cutting the population over time will be the “smart, green” thing to do. Supporting wage labour, or the programs to support former wage labourers, will become a burden to the countries that try to do so vs. the countries that don’t care about generous social programs, the same way the Confederacy and its slave economy couldn’t keep up with the Union’s transition to wage labourers

    • Randy M says:

      I think the ideal case scenario is a UBI coupled with a rise in creative enterprises allowing money to be passed basically for status among a community. Everyone will be a writer, or artist, or game designer or professional athlete and/or consumer of such stuff, and vote with their excess funds for which they prefer. Think Youtube writ large.

      Possibly a lot of this will be deception behind the scenes just to keep people content and “productive.” But that’s not too bad compared to possible alternatives.

    • Mabuse7 says:

      People keep talking about a UBI, but I think we can do better than that. People want to feel independent, and no matter how generous or unconditional it is, a regular cheque from the government is going to feel like dependence. I think we should look to ancient Sparta for inspiration; Sparta had something analogous to the fully-automated economy we’re talking about thanks to their massive population of Helot slaves, and the Spartan government distributed this bounty to its citizens not by paying them regularly, but by giving each male Spartan an estate and an allotment of slaves when they reached adulthood. This government grant of wealth was designed to give each male Spartan the means to never have to work, so they could spend all of their time training for war; hence Sparta’s large professional army. The granted wealth was kept officially and administratively separate from a citizen’s other holdings, and was returned to the government when the citizen died, so the system was sustainable as long as the land:citizen and slave:citizen ratios remained stable (through either population control or conquest, usually conquest). I think that with modern financial techniques we could easily develop a system where every citizen gets a grant of capital in a universal index fund when they reach adulthood, and in the roboticised economy of the future a grant large enough for someone to live prosperously on would be trivially small compared to the resources a government could bring to bear on the problem.

  10. Wrong Species says:

    It’s bugging me that you’re comparing total number of manufacturing jobs with percent of men in the labor force. It should either be both percent or both total.

    Consider the surprising complementarities between information technology and employment in banking, specifically the experience with automated teller machines (ATMs) and bank tellers documented by Bessen (2015). ATMs were introduced in the 1970s, and their numbers in the US economy quadrupled from approximately 100,000 to 400,000 between 1995 and 2010. One might naturally assume that these machines had all but eliminated bank tellers in that interval. But US bank teller employment actually rose modestly from 500,000 to approximately 550,000 over the 30-year period from 1980 to 2010 (although given the growth in the labor force in this time interval, these numbers do imply that bank tellers declined as a share of overall US employment). With the growth of ATMs, what are all of these tellers doing?

    I find it odd how people just gloss over that. The right story is the simple one, where ATMs reduced employment among tellers. Rise through population growth doesn’t count.

    • it lends evidence against the argument that new technologies completely eliminate jobs. the old jobs don’t completely go away but rather are surpassed by the creation of new jobs.

      • Wrong Species says:

        Which narrative do you think better explains the effects of the industrial revolution on farming:

        Machinery is a substitute for labor, gradually reducing the need for laborers in agriculture. You can see this in the declining percent of Americans employed in the industry.

        Machinery is a complement to labor, allowing farmers to specialize in a few tasks instead of the myriad of tasks they used to do. The evidence for this is the fact that there are still jobs for farmers.

  11. Peter Gerdes says:

    As long as we have a capitalistic economy, the vast majority of people don’t inherit substantial capital and don’t switch to something like UBI that gives an alternative means to retain one’s social status (and access to rivalrous goods like the most prestigious schools, poshest/safest neighborhoods, newest gadgets etc..) **of course** automation won’t result in unemployment as someone is getting all those resources to distribute and will hire the other people.

    Maybe humans won’t be as efficient as robots or as good at their jobs but human nature being what it is the limited supply of individuals and the psychological attraction of interacting with other humans means whoever is making the money will hire people. Even though you get equal sound quality from recorded music people continue to attend and hire live performers and as automation makes materiel goods dirt cheap they will poor more and more of their money into statusful consumption of goods made by humans.

    This might conjure up images of human chess pieces, prostitution and servants but in fact its much more likely to take the form of consuming art, or even ice cream made by hand. Indeed, as machines have become better at matching beats than humans preferences shifted and people began to pay more for DJing with mistakes that demonstrated authenticity. It will also take the form of things like TeD talks where high status people get to exhibit their status by associating with statusful academics or sending their children to super elite universities. This second level of elites will in turn want their own similar, if slightly cheaper, version of human made products and so on.

    In short its the fundamental scarcity of human labor and our innate psychological drives to relate to and be complimented by other humans that ensures the underlying demand for human labor…at least until we start wireheading and changing those things.

    • Wrong Species says:

      Surely there is a limit to this. I might signal my sophistication by buying crafts made by a person but is that enough to sustain tens of millions of jobs?

    • Bugmaster says:

      I may be a special case, but personally I much prefer to interact with robots. For example, I used to visit electronics stores quite a lot when I was younger; nowadays, I only go there if I have some hardware emergency and can’t wait for shipping, or if I want to physically handle some product before buying it (e.g. a mouse or a keyboard). Going back to the days of in-person retail, when I’d need to spend 15 minutes convincing the clerk that motherboards are unlikely to be found in the TV department, then waiting another 15 minutes as his buddy tries to upsell me on something completely irrelevant… yep, I just checked, still feels like hell.

      • vV_Vv says:

        Similarly, I prefer Uber to taxis even where they cost more or less the same because with Uber I don’t have to explain to the driver where I’m going, then pay with cash or physical credit card while making sure he’s not trying to scam me.

      • Jiro says:

        If upselling is profitable for the company, then companies will have their robots try to upsell to you as soon as robots are advanced enough that they can do that.

    • sharper13 says:

      Either AI+robots are cheap, in which case you don’t need to “inherit substantial capital” to have some yourself, or else they are expensive, in which case they can’t replace everyone.

      You can postulate scenarios for both situations, but it appears to me to be a contradiction to postulate both “you must be rich to own any” and “they are everywhere, replacing everyone” at the same time.

      • vV_Vv says:

        Either AI+robots are cheap, in which case you don’t need to “inherit substantial capital” to have some yourself, or else they are expensive, in which case they can’t replace everyone.

        Suppose you are a supermarket cashier and at some point you are made redundant by a self-checkout machine. Can you buy a self-checkout machine yourself and rent it to the supermarket to maintain your income?

        If you are a redundant car factory worker can you buy an assembly robot and rent it to the factory?

        Capital tends to concentrate as long as it can be used productively, redistribution of capital usually happens due to mass violence or disease.

        • sharper13 says:

          Well, the first thing I’d ask is “Why would I want to?” Most supermarket cashiers don’t aspire to rent machines to supermarkets and most car factory workers don’t aspire to rent assembly robots to a factory. That’s because they don’t have a comparative advantage in the skills relevant to the renting machines occupations and most of those places aren’t going to “rent” anyway.

          What you can do in either situation is go work on something you have a comparative advantage in and then purchase more of the now cheaper groceries or automobiles made possible by the automation.

          Can a ditch digger made redundant by a backhoe purchase a backhoe and rent it out to a farmer? No, but that hasn’t stopped those same people from getting a different job and people enjoying more variety and quantity of food from the farmers.

          As the capital to labor ratio moves towards more capital in use and higher productivity, it may affect people unevenly, but everyone eventually gets wealthier than they would have been. Or are you arguing we should ban farm machinery so that more people have to have manual farm jobs instead of whatever else they are currently doing?

          • MugaSofer says:

            What you can do in either situation is go work on something you have a comparative advantage in and then purchase more of the now cheaper groceries or automobiles made possible by the automation.

            Can a ditch digger made redundant by a backhoe purchase a backhoe and rent it out to a farmer? No, but that hasn’t stopped those same people from getting a different job and people enjoying more variety and quantity of food from the farmers.

            1. You seem to have gone from arguing that technological unemployment is fine because people could just purchase capital, to arguing that technological unemployment is impossible.

            2. You’re quite right that in real life the workers put out of a job by automation don’t have comparative advantage at purchasing the machines that put them out of work. Corporations take advantage of economies of scale, and have access to better automation experts. So … why would this cease to be true in a world with widespread technological unemployment?

  12. Steve Sailer says:

    One thing to keep in mind is that America has a lot of square feet of housing per capita these days, so there is a lot of room under a roof somewhere for people who might not be working as hard as they could.

  13. bodo says:

    Small point, you write:
    “In 1970, educated and uneducated men were about equally likely to be PAMLFNPers. The rate for educated men didn’t change. The rate for uneducated men plummeted.”

    But PAMLFNPers are the nonparticipants. So that is not plummeting but rising. The PAMLFPR is plummeting for uneducated men.

    • Yorwba says:

      There are a few similar mix-ups at other places in the text:

      “decline of manufacturing = increasing disability claims = decline in PAMLFNP”

      “3. Or there was some mysterious factor causing PAMLFPR rates to rise from 1960 to 2000”

      “Third, it is very hard to find a temporal correlation between the apparent effects of automation on manufacturing, and the rise in PAMLFPR. PAMLFPR has been rising very steadily since the 1960s”

      That’s the problem with long acronyms that are mostly identical, but mean completely different things. They make it so easy to end up saying the exact opposite of what you intended to say.

  14. tentor says:

    > Frickin’ Germany, always making everyone else look bad.
    That Germany is an outlier in this chart is probably more related to the end of the GDR than any other differences. Would be important to know if GDR territories are included in the 1990 number.

  15. BobCatP says:

    But maybe this is better viewed as a race between two competing forces; generational churn producing students with the right set of skills, and technology making new skills obsolete. I don’t know why this should have increased recently, but it seems like – at least for the middle class – this is a race they are now losing.

    Maybe this has something to do with the pace of change and timing of ‘waves of innovation’. I’m not sure how well supported the whole waves of innovation idea is, but essentially whereas step-changes in technology used to take lifetimes or generations, they now appear to happen increasingly often, so there may be huge step changes in technology during the lifetime of a career. A culture of ‘do almost all your studying before the age of 25’ means that when people turn 45 and their job is automated, they don’t have the ability to easily retrain.

    • azhdahak says:

      A culture of “do almost all your studying on your own dime”, you mean.

      I was talking to an Uber driver a while ago who told me about a problem a lot of people in her generation are having right now: they have kids, and have to help those kids out with the costs of college, but they also have to pick up the costs of their own retraining.

      My first job was with a company that developed all their technological tools in-house, so they had to train people. And, for whatever reason, they refused to document anything — they theoretically had an intranet wiki, but it was next to impossible to find what you were looking for on it, and the documentation was hopelessly incomplete or out of date on the rare occasion that it existed. But there still wasn’t very much friction over it. The training got done.

      • ksvanhorn says:

        “for whatever reason, they refused to document anything”

        As someone who been in the software industry for 30 years, my experience has been that this is the rule rather than the exception.

  16. Alethenous says:

    Please don’t apologise for posts being long-winded. That’s one of the reasons I’m here. Brevity may be the soul of wit, but not everything is a punchline.

    • When one removes the pictures, it’s not that long anyway. If we cannot read 3,000 words in one sitting than that is an indictment of our own limitations if anything.

      • Jiro says:

        That’s only true if the pictures are inessential to the post. This may be true, but it isn’t something I can know a priori.

      • AG says:

        3000 words is just a regular ol’ one-shot, consumed in minutes, by fanfic standards. 😛

        • The Nybbler says:

          Since when does fanfic have standards?

          • AG says:

            Since the time the author/s of the Vulgate cycle built their Mary Sue OC Galahad specifically to two-up de Troyes, out-Pure-ing Percival AND Lancelot bashing with one stone.

      • anonymousskimmer says:

        Some of us are limited.

        I got through it, it was interesting enough to keep my attention, but it would have put me to sleep if I wasn’t interested. As is, I believe I skimmed a few spots.

    • Mark Bahner says:

      Please don’t apologise for posts being long-winded. That’s one of the reasons I’m here. Brevity may be the soul of wit, but not everything is a punchline.

      Yes, I haven’t even come close to reading the whole post, but what I’ve read so far is gold! Lots and lots of good stuff. The only reason to apologize would be long-winded bad stuff. 🙂

    • StellaAthena says:

      I feel obligated to point out that brevity is not the soul of wit, and the character in Hamlet who claims as much is consistently portrayed as an idiot who thinks he’s a genius.

  17. JovialJellyfish says:

    [Unrelated; feel free to delete when read, but I don’t know how to contact you privately.]

    There’s a broken link: the text “Part of Amazon Affiliate Program” at the very bottom of the page (past the comments section) links to https://slatestarcodex.com/amazon, which no longer exists.

  18. Tracy W says:

    How undignified a job should people be forced to do?

    As someone who has worked as a nurse’s aide, cleaning up piss and shit, I don’t get this in the context of health sector jobs. I mean, the job was sometimes very unpleasant, I do have a sense of smell. But if someone is that disabled or sick someone has to help and clean up afterwards. You are helping people every day. And you frequently have someone to talk to.

    Wouldn’t a really undignified job be selling something you didn’t believe in? Or having to regularly implement policies you think are terrible?

    • JovialJellyfish says:

      Many people – I, for instance – don’t think like that. Apparently you do. People have different ideas about what is undignified.

      • there is a subjective element to it. some used to see sideshow and other carnival jobs as undignified, but the performers didn’t.

      • Tracy W says:

        Genuine questions: So, if you ever are so sick that you can’t manage to use a toilet without help, or perhaps your parents are in that state, what do you think should happen? Should you or your mum be left to stew in your own wastes? Or do you really think the hypothetical person helping you would be inherently undignified?

        • albatross11 says:

          “Undignified” means low status, not useless or unimportant. Plenty of jobs that absolutely must be done for our society to continue to function have low status–for example, prison guards, janitors, and garbage collectors. Many high status jobs could probably go away with little damage done to our society–for example, most plastic surgeons, sports start, and media celebrities.

          Taking a job with lower status is very much like a pay cut, or like taking a job with worse working conditions–people forced by circumstances to do so will feel rightly like their lives have gotten harder, and will try to avoid doing so if they can.

          • Tracy W says:

            “Undignified” means low status,

            Does it? Then why not say “low-status”? And, since when was working on an assembly line as unskilled labour “high-status”?

          • Futhington says:

            @Tracy W

            > “Then why not say “low status”

            Because it seems like Scott’s trying to use somewhat natural language? Outside of the SSCsphere nobody would say “I won’t do that because it is low status”, they’d talk about how it’s undignified or shameful.

            Anyway, I’m going to disagree with the poster above: the job is considered undignified because it’s working with filth. Same reason jobs working with waste have always been considered pretty undignified, and sometimes outright spiritually tainted. People don’t like interacting with shit.

            Further to that: while the job of, say, a sewage technician might involve a lot of filth it’s A. Sufficiently removed from it and B. Sounds like skilled, qualified work. With things like mothering (which you mentioned elsewhere) are probably not considered undignified because the focus is on actually rearing children rather than just cleaning up filth.

            Finally I’d argue it’s considered undignified because it’s not something it seems like anyone would *want* to do. Same reason fast food is undignified basically; it doesn’t seem like cleaning up after incontinent old people is something to aspire to. Whereas being a good mother, ensuring people have safe and clean water, hell even just being a farm worker all seem not only socially useful (actual social usefulness notwithstanding) but something somebody might conceivably desire to do. Nobody wants to, or at least I and probably many others believe nobody wants to, clean up old people’s shit for the rest of their life.

          • Tracy W says:

            @Futherington:

            Outside of the SSCsphere nobody would say “I won’t do that because it is low status”, they’d talk about how it’s undignified or shameful.

            A. This is inside the SSC sphere. The bit I originally quoted was Scott’s own direct words, not Scott modelling the mind of a 50-something laid off manufacturing worker.

            B. Can we seriously ask: “How low-status a job should people be forced to do?” How could any society possibly ever only offer middle- to high-status jobs? It’s a relative scale for heaven’s sake!

            the job is considered undignified because it’s working with filth.

            That may be. That doesn’t mean that’s sensible on the part of society. Having a prejudice against people who clean up filth is silly – and not what I’d expect of a trained doctor.

            These sorts of attitudes aren’t set in stone. There was a public prejudice against sodomy for a very long time, it was even illegal, and of course sodomy also involves cleaning up filth (well at least one hopes!) But that’s changed a lot now.

            Nobody wants to … clean up old people’s shit for the rest of their life

            Nobody wants to tighten bolts on car doors 8 hours a day, 40 hours a week, for the rest of their lives.

          • Nornagest says:

            How could any society possibly ever only offer middle- to high-status jobs?

            Smartassed answer: if everyone’s a subsistence farmer, then everyone’s got a middle-status job.

          • albatross11 says:

            I’m assuming low-status jobs require some compensation (better pay, benefits, hours, job security) for a given worker to take them. On one end, that can mean that a job that most people could do but most don’t want to gets decent pay (garbage collectors), but that’s not a guarantee, it depends on the bargaining position of the people taking the jobs.

        • JovialJellyfish says:

          I should have been clearer. “Undignified” varies from person to person. While I consider it undignified to perform people’s bodily functions for them, I know first-hand that people exist who actively enjoy the jobs that require it.

          • Tracy W says:

            I’m afraid I still don’t get it. So, if you or a loved one needed someone to help toilet you, you would seriously regard said hypothetical helper as undignified?

            Is it that you think of ‘dignified’ as a synonym for ‘actively enjoy’?

    • Scott Alexander says:

      Thanks, you’re right that was wrongly phrased. Deleted.

      • Aapje says:

        You should just have called it a ‘shit job.’ No offense can be taken if people take it literally.

      • Tracy W says:

        I wasn’t offended. And I know it’s a common view. I just don’t get it as a view.

        • Shannon Alther says:

          I just don’t get it as a view.

          It’s tempting to think that working directly with filth is what makes people view these sorts of jobs as undignified, but that doesn’t hold water. Non-laparoscopic surgery, wastewater treatment, and a lot of farm work and veterinary medicine are at least gross on the same order of magnitude, but not all are of equally low status.

          It looks like the common factor is (apparent) simplicity. The average person working in a different field wouldn’t have the slightest idea how to begin practicing law, programming a computer, or performing surgery. Interior decoration, painting, and writing aren’t mysterious, but there are clear lines between the skill of the untrained, the amateur, and the professional.

          But nursing aide, garbage collector, etc. aren’t particularly cool, are somewhat unappealing (to put it mildly) and provoke a snap judgement of “Anyone could do that.”

          • Tracy W says:

            Mothering (gendering is deliberate) isn’t particularly cool, involves regularly dealing with shit and other gross stuff, and “anyone” can do that. But it’s not regarded as undignified.

          • Shannon Alther says:

            I’m tempted to call ‘mothering’ the exception that proves the rule. Even so, many people experience anxiety over the difficulties of being a parent and sympathize with new parents over their struggles; parenthood is something that many if not most people aspire to; and dealing with soiled diapers and the like is a small part of parenthood, greatly outweighed by the tasks and time that don’t involve “shit and other gross stuff”, and rewarded with the plethora of things that people consider the rewards of parenthood, rather than cash.

          • Tracy W says:

            @Shannon: I’m a mother, and in my experience so far there’s a lot of gross stuff. They’re finally toilet trained and they discover gross-out jokes. Not to mention the stomach bugs.

            Obviously mothering has emotional rewards, but then so does nurse aiding, at least in my experience working at a centre with a variety of patients. I guess it may well be different if you have one patient every shift and they’re a pain.

          • Shannon Alther says:

            @Tracy W

            As both a mother and a nursing aide, you have an inside perspective on exactly how much exposure to biohazardous matter each requires. But nursing aides aren’t the final arbiters of what a society (or parts of a society) thinks is an undignified career. Take the outside view by putting on your Random Observer hat and thinking about garbage collectors. They deal with trash all day, it’s not a very complicated or demanding job, and it doesn’t pay very well. That’s the formula for low status.

            Now, in many cities garbage collectors are well compensated, belong to powerful unions, and the labour is less gross than it is physically demanding (although trash can also be dangerous). But the Random Observer doesn’t know that. These things are also true of many nursing tasks.

            But they aren’t true of motherhood. Our Random Observer probably doesn’t fully associate parenthood with childcare, and when they do the first thing that springs to mind is probably not how overwhelmingly onerous dealing with sick children is. As you yourself noticed, your children are now toilet-trained, which means that ideally you’ll have at least another half-century with them (another decade or two in a supervisory capacity, maybe?) where you don’t need to clean up after their bowel movements or flu episodes.

            TL;DR – low status happens when the average uninformed person reasons thus: “That’s an easy, low-paying, gross job.” Most people don’t feel this way about motherhood, probably in large part because just about everyone is either a mother or has a very intimate relationship with one.

          • Tracy W says:

            @Shannon: Random Observer doesn’t reason “I might get sick, or my mother might get sick, and need someone to clean her up so she doesn’t stew in her own body wastes. Therefore I respect people who do this job.”? (Ditto for garbage collectors.)

            I mean, this is not complex reasoning in a society that at least pays lip service to the dignity of labour and fundamental human equality. I know a lot of people don’t take this step, but I still don’t get why.

          • MugaSofer says:

            The fact that a job is important and helps people doesn’t factor in much to its status.

            Unless you’re believed to be doing it for charitable prosocial reasons, rather than because you really need the money (which means either you’re literally working for a charity, or it’s a “heroic” profession like firefighting or the military).

    • Steve Sailer says:

      Thanks for doing those jobs.

    • Deiseach says:

      It’s very necessary work, but it’s considered unskilled or semi-skilled and low status*, therefore it ‘should’ be low-paid (because why are you gonig to pay someone top dollar for shovelling shit, they’re so dumb they can’t get a better job, they don’t deserve higher wages).

      That is part of the problem: the old economy is changing to service economy, and service jobs are still regarded as not quite ‘real’ jobs, of being low-status, of being something anyone can do so why pay more for a trained monkey? Men who are laid off then have to look for work in the kinds of service professions that women have worked face the low wages there for the combination of those reasons: ‘care-taking’ is ‘women’s work’ and women’s work is lower paid (for all the reasons, I’m not rehashing the wage gap argument).

      *Anybody can change diapers; this is, ironically I think, because such care-taking used to be part of the unpaid labour within the home done by women and so it is still tainted with the stigma of “Aunt Sally the spinster daughter living at home with her parents used to do this for Grandma Josie when Grandma took to her bed for free, why pay a care home worker $$$$$$ to do it for mom and dad now?” See the discussion over paying childcare workers that we had on here before, with the counter-argument being “anyone can be a parent and look after their kids themselves without training, ergo childcare workers in nurseries and creches don’t need that training, so they should not be paid that level of wages and this would make childcare affordable so women could go out to work!” Not much recognition that some women would be subsidised in ‘real’ jobs by other women working for less.

      • Tracy W says:

        It is low-skill work. But it is labour intensive. And nursing homes are very expensive as it is. A rise in wages for care workers relative to the median wage would make nursing homes even more unaffordable for the general population.

        Which would be rather miserable for those who can’t manage their own toileting.

      • nadbor says:

        I don’t get it. My mum always told me that prices in a competitive market are determined by supply and demand. You seem to be saying that they’re determined by how much people *think* things are supposed to cost. This can’t be literally true and I’m struggling to find an interpretation that would render it even approximately true. Care to explain?

        • The Big Red Scary says:

          Perhaps the above considerations affect supply more than demand, since people looking for work are trying to optimize many things in addition to their income. For me, actually, income goes into the constraints rather than into the utility function of the optimization problem.

          But I suppose it also affects the demand size. Surely the very rich think their shit doesn’t stink and are both more willing and more able to pay someone more to clean it up!

        • Thomas Jørgensen says:

          Markets in labor are not even remotely efficient. And prestige does influence how much of a wage you can command, sometimes with a total disconnect to how much value your labor is worth. – There are many jobs in finance which can be demonstrated to on net *destroy* wealth for the people employing those workers, and which still command very high pay. Because the world is mad.

          • nadbor says:

            I think you may be misunderstanding the whole ‘supply and demand’ thing. Prices are not determined by any ‘true value’ only by the value to the buyer and the seller – defined as the amount of money they are willing to pay or accept for the thing. If financial companies value the services of a professional prestigious wealth-destroyer at 500k then wealth destroyers will be able to charge up to 500k. But only if they are scarce. If more wealth-destroyers show up than there are companies willing to pay the 500k then they will have to either accept lower wages or find another industry to bamboozle, no matter how prestigious everyone thinks they are.

            If you’re the only babysitter in town you will make as much money as the richest family is thinks it’s worth to them to pay to have a night out alone. If literally everyone in your town is a lawyer, sweeping floors will pay more than lawyering.

        • Tatterdemalion says:

          Consider an economy where there are a bunch companies who want to employ widget-makers and a bunch of widget-makers looking for work. The value of the widget-makers time to them – that is, the level at which they would accept a job if the choice was that or nothing – is $10 an hour; the amount of value employing a widget-maker supplies to a company is $20 an hour.

          Economics 101 tells us that the hourly wage for a widget maker will be between $10 and $20. But there are a lot of other factors, not just supply and demand, that will effect how that range where some value is being created for both sides is carved up, and “how much people think things are supposed to cost” will be a big part of it.

          • nadbor says:

            Thats not what economics 101 says. Both econ 101 and common sense tell us that way more people will be willing to make widgets for $20/h than for $10/h but a lot more companies will be willing to employ a widget maker for $10/h than for $20/h. The equilibrium price is determined as the one that makes quantity supplied equal quantity demanded.

            There is no process separate from supply and demand that determines the exact price between $10/h and $20/h. It’s all down to how much people are willing to pay and how much other people are willing to accept (aka ‘supply and demand’).

            Preconceptions about the ‘right’ level of prices may enter the equation, but only indirectly, via supply and demand. Someone moves to a more expensive city and refuses to buy items they deem outrageously expensive, but then they gradually adjust and thinkg nothing of it. And there are countless other similar biases.

            I tried to figure out how this would work in the context of Deiseach’s comment on care workers. Are men less willing to pay for childcare because they regard it “women’s work” so they’d rather do it themselves? Are women more willing to accept lower pay for it than they would for other kinds of work because they’ve been brainwashed into liking it? Sounds rather far-fetched to me and anyway it’s probably not what Deiseach meant, hence my question.

            I’m still waiting for anyone to show any plausible mechanism leading from people *thinking* that childcare should be cheap to it being cheap.

            Everyone thinks rent is too damn high but as long as new apartments don’t pop into existence from thinking alone, that is doing zilch to keep rents down.

            On the other hand, most people I talk to agree that clothing has gotten ridiculously cheap compared to what we remember from back in the day but our thinking is again not enough to drive up the price in the face of abundance.

      • caryatis says:

        >See the discussion over paying childcare workers that we had on here before, with the counter-argument being “anyone can be a parent and look after their kids themselves without training, ergo childcare workers in nurseries and creches don’t need that training, so they should not be paid that level of wages and this would make childcare affordable so women could go out to work!” Not much recognition that some women would be subsidised in ‘real’ jobs by other women working for less.

        So what’s the problem with higher-wage women hiring lower-wage women? I wasn’t privy to some of this discussion, but I don’t see why that would be bad. And it doesn’t seem to be a criticism that gets directed at higher-wage men who outsource traditionally male jobs like car repair or lawn care.

        • The Big Red Scary says:

          I didn’t follow that discussion, but if two families, one rich and the other poor, both have babies, and the poor family has to send their baby to a crowded nursery while they take care of the rich baby, I at least can’t help feeling that this is a tragedy. You can say all you want about comparative advantage and the freedom of the market, but it sticks in my craw.

          A solution, used by some friends: they hired a nanny with a baby to come
          over to their house with her baby. This way the nanny could spend time with her baby, and the babies could play together. Our solution: my wife and I both cut back on work, though my wife much more than I since she was nursing. But we could afford to do so.

    • Mark Bahner says:

      “Wouldn’t a really undignified job be selling something you didn’t believe in? Or having to regularly implement policies you think are terrible?”

      I’ll leave names out, but there was a situation in Florida where a house had all the toilets and drains backing up. A person went outside in the Florida sun and dug up the drain field. The field was basically clay pipes with gaps between them, and roots grew into the gaps and clogged the pipes. This person worked for hours in the sun and unbelievable stench. I wanted to help, but I just wasn’t strong enough…my gagging reflex was stronger than I was. (If I’d thought about it, I might have gone and purchased a respirator, and maybe I could have helped some.)

      In the end, the new drain field was put in, and all the toilets flushed and the sinks drained. To me, the job wasn’t “undignified.” It was heroic.

  19. azhdahak says:

    I’m not an economist and can’t contribute to the economic discussion, but I can say something about working unskilled jobs, since that’s what I do.

    At least where I live, there are plenty of these unskilled jobs. The problem is that wages are low and costs are high. The average rent around here is $900/mo, and that’s with roommates. You can’t go below $700/mo for a room without placing yourself in serious physical danger.

    After taxes, I end up taking home about $700 per biweekly paycheck. So over half of my income ends up going to rent and bills, and there’s no way around that other than to earn more money. After food, transportation costs, and the occasional visit to a health clinic about job-related chronic pain, I can’t really build up savings. If I didn’t have to work — if I could get by on benefits or sponge off my parents — why would I?

    And I see a lot of people making that decision. My employer hires essentially everyone who applies and can pass a drug test. Over half of new hires quit or get fired within two months.

    Some of them are college students or kids living with their parents, but some of them aren’t — and they aren’t all quitting. A lot of people get fired for not bothering to show up, or for showing up drunk or high, or for storing several hundred dollars of weed at work in an employer-provided container with their name on it and stinking up the place, or for fucking in a meat locker. I am not making any of these things up.

    I would imagine that people who get fired for fucking in a meat locker or showing up visibly high would have a hard time holding down a job anywhere. What are they going to do? They just aren’t functional enough to work. I don’t know if the problem of people just not being functional enough to hold a job has gotten worse over the last few decades, but that’s one place I’d look.

    • benwave says:

      I remember reading a study where parent were given small fines for delivering children late to kindergarten, in a society where previously this was enforced mostly by social shame. Tardiness went up, people paid the fines. Then they stopped the experiment, and things didn’t go back to normal – tardiness stayed high, the system of social shame no longer operated.

      When work is just seen as a transaction rather than a relationship, it’s not really surprising that people don’t respect it and do things like show up drunk.

    • I think in such circumstances, one can make a rational case to not work.

    • Deiseach says:

      I don’t know if the problem of people just not being functional enough to hold a job has gotten worse over the last few decades, but that’s one place I’d look.

      Look at the lyrics to the 1964 Roger Miller song “King of the Road”:

      Trailer for sale or rent, rooms to let, fifty cents.
      No phone, no pool, no pets, I ain’t got no cigarettes
      Ah, but, two hours of pushin’ broom
      Buys an eight by twelve four-bit room
      I’m a man of means by no means, king of the road.

      Plainly this is some cheap flophouse, but at the same time – two hours of menial labour will get the price of a room for the night. People who were not able to hold down a full time job could scrape by doing the bare minimum because there were enough cheap, casual jobs and even cheaper prices out there. Can we say the same today – that two hours of under minimum wage labour will pay for a night’s room?

      • Tracy W says:

        Zoning has a lot to answer for.

        • albatross11 says:

          I’m curious how much effect the rising floor of acceptable living conditions (no flophouses, required indoor plumbing, whatever) has had on employment. My guess is that the main impact is on the currently homeless–some of them might have a place to sleep at night if they could rent a small room somewhere for almost nothing.

      • benwave says:

        That’s a good point. Similarly, single incomes used to be enough to support a whole family – spouse and children included. Cost disease has a lot to answer for!

        • Cliff says:

          I’m not sure that it’s not still enough. People just consume a lot more. Houses are much bigger for example

          • baconbits9 says:

            Single income family here. We looked at a lot of houses that people in our social demographic consider ‘affordable’ and decided that they weren’t for us, our cars are 12 and 17 years old and we don’t pay for cable TV and a handful of other commonly used luxury items.

          • albatross11 says:

            baconbits:

            We’re in the same boat–we live in a fairly old, small townhouse, whereas most of our friends live in big single-family houses. This is the result of being a one-income family (my wife deciding to stay home with the kids) and wanting to limit the commute time (which tied in with keeping the kids in their Catholic school, staying in a parish we love, etc.). I think this was overall the best choice for us, but it has some very visible costs in terms of having enough room to have people over and such.

      • INH5 says:

        Plainly this is some cheap flophouse, but at the same time – two hours of menial labour will get the price of a room for the night. People who were not able to hold down a full time job could scrape by doing the bare minimum because there were enough cheap, casual jobs and even cheaper prices out there. Can we say the same today – that two hours of under minimum wage labour will pay for a night’s room?

        In some cities you can. In the Los Angeles metro area there are, as of this writing, 64 Airbnbs available for a stay from March 23 through April 23 at rates of less than $20 per night, in a state where the minimum wage is $10.50 per hour.

        Even in some cities that are notorious for high housing costs, you can get pretty close. In the San Francisco metro area there are 19 Airbnbs available for the same period of time at rates of less than $25 per night.

        And speaking as someone who spent several months living in various Airbnbs a couple years back when I was in a tight spot financially, I would expect most of those places to be a lot more comfortable than the flophouse described in that song.

        Now, whether you could actually find a job that would pay you minimum wage and allow you to work just 2-3 hours per day but wouldn’t impose extra costs (gas and car maintenance when working as a Uber driver, for example), I have no idea, because I’ve never lived in either of those cities.

        • anonymousskimmer says:

          Now, whether you could actually find a job that would pay you minimum wage and allow you to work just 2-3 hours per day but wouldn’t impose extra costs

          Package handler – FedEx/UPS

    • neonwattagelimit says:

      It seems to me that there’s gotta be a sort of self-reinforcing cycle going on here. Like, if your job doesn’t pay you enough to do anything but stay alive and keep working, you’re not going to have much hope for the future. If you can’t build wealth, you can’t invest in your skills, you can’t even go on vacation – why wouldn’t you show up high or fuck in a meat locker? What’s the point in exercising self-control if the future is just grinding poverty?

      Then, of course, the job will wind up attracting the sorts of people who are more likely to show up and high and all that, because anyone with better options will do something else. And so the cycle continues.

    • angularangel says:

      I don’t know if the problem of people just not being functional enough to hold a job has gotten worse over the last few decades, but that’s one place I’d look.

      Hmm. I forget, has the connection between lead pollution and poor impulse control been borne out by further research? If so, then that might actually be a significant contributor, assuming enough people are still affected.

  20. outis says:

    It seems clear that, so far, the impact of automation on unemployment/underemployment has been dwarfed by that of globalization and delocalization. But that is precisely why there is such an intense focus on technological unemployment: we are hungry for arguments to justify our policy choices of the past ~20 years. I’ve seen articles that explicitly make the argument “the jobs might have gone to China, but it doesn’t matter, because now they’re going to go to robots anyway”.

    In that respect, I would be interested in a closer look at the condition of American manufacturing. On one hand, whatever indicator economists like is up; but on the other hand, Apple says that they could not possibly make the iPhone in the US, because the manufacturing capability is simply not there anymore. What’s going on there? One simple observation is that our production may have gone up, but maybe our consumption has gone up even more. Has American manufacturing’s share of the products we actually consume gone up or down?

    • Scott Alexander says:

      That doesn’t seem clear to me at all (not saying it’s false, just not clear).

      The Ball State paper estimated that manufacturing job losses were 13% globalization, 87% automation
      Autor et al estimated about 25% globalization, 75% automation

      I am suspicious of both (someone said Ball State’s model was wrong, and Autor is looking specifically at China-related globalization only), but in the face of this kind of stuff I feel like “globalization is more important than automation” is a controversial position you need to defend.

    • postmodernists and accelerationists call it deterritorialization. immigrants move to America; Americans must move to find new jobs . both the immigrant and the native have to move…a constant state of flux

      • christhenottopher says:

        Quick thing here, internal migration in the US reached it’s peak back in the early 80s and has actually been falling. Increasing immigration doesn’t really seem related to amounts of Americans moving around inside the country so linking these two concepts doesn’t seem really useful to me.

    • sharper13 says:

      Comparative advantage… I think you’re confusing the different between capability and cost.

      Apple can’t make the iPhone in the US for the same price point as they can make it elsewhere, because for small hand-laborious tasks which may violate US regulations in some ways (labor, environmental, etc…) other places can make an iPhone for less.

      I don’t think anyone is suggesting it would be impossible to make an iPhone in the US, it’s not like the factories they’re using right now preexisted their desire to make iPhones, so they could have built new ones in the US as well. Apple means that they couldn’t make and sell nearly as many of them. A quick Google came up with one estimate that the price could double.

      So if no one could make smartphones for cheap elsewhere, than some people would pay the double-price and Apple would still sell iPhones, just not nearly as many of them. Of course, if they were the only ones forced to manufacturer in the US, then their competition would take even more of their smartphone market share and it might become uneconomical for them to be able to sell any of them, but that’s due to cost, not capability.

      • Thomas Jørgensen says:

        The iphone is a terrible example, due to the very low cost component of the hardware, and the high profit margins. Iphones *could* be manufactured and sold at current prices in the US. It would not even make that much of a dent in their profit margin.

  21. benwave says:

    So I have some thoughts on this topic.

    The first of which is – I have a vague feeling that it is generally agreed that the last hundred or so years, give or take, have been an uncharacteristically wealthy and egalitarian era. How much of the feeling that the fruits of human labour are (some combination of) getting smaller and being more concentrated at the top can be explained simply by some kind of regression to the mean? I find it plausible that some combination of technology and organisational forms teamed up around the turn of the century to deliver rapidly expanding wealth and increasingly widespread distribution of wealth to an unusual degree; but the technologies and organisational forms which win the competitions that we have collectively decided to run have stopped being so unusually good at that as time has gone on.

    The second thing is – conditional on the above conclusions being correct, then technological unemployment is not to blame for increasing concentration of wealth and the step backwards in real disposable incomes in the developed world recently (I recognise that in other countries such as China, pretty much everything is better for pretty much everyone than in 1900). I actually find it sort of encouraging that the march of technology is not to blame for this, because I actually rather like the march of technology!

    • angularangel says:

      Huh – Heres an interesting question. How closely does this egalitarian era map to the rise and fall of the Soviet Union and international communism as a force to threaten the wealthy classes? Perhaps when communist revolution was something to be taken seriously, it was felt necessary to embrace measures that would insure the health and moderate wealth of the middle class as a buffer against revolution, and now that the threat has passed, this is no longer so necessary? Though it seems unlikely that something so large could be coordinated without the cause and effect being commonly known. :/

      • Swami says:

        Actually it does map the fall of communism but for different reasons. With the fall of socialist beliefs, we see the introduction of a billion plus one time serfs entering the market, increasing competitive pressures on unskilled labor and increasing returns to capital as entrepreneurs scrambled to find productive uses for a billion human beings.

        Net result has been the greatest escape from poverty in history, the fastest growth in median global income ever, unusually high returns to skilled labor and capital and slower than previous level growth in developed nation lower skilled wage rates.

        It’s not so much a top down thing, rather a bottoms up shake out due to the radically differing scarcities of capital, skill and unskilled labor. By the way, the change is now over. There are no more new China’s and Indias sitting on the global sideline.

    • Cliff says:

      “in other countries such as China, pretty much everything is better for pretty much everyone than in 1900”

      In what country is that not true?

      “the feeling that the fruits of human labour are (some combination of) getting smaller and being more concentrated at the top”

      Maybe the feeling is the result of inaccurate media characterizations and partisanship?

      • benwave says:

        from 1900 sure, but as others have pointed out increasing polarisation has meant that a significant proportion of people in for example America are not better off than, say, that percentile person would have been in about 1970-80.

        On the fruits of human labour being more concentrated at the top, I would honestly be surprised if you were to argue against this. Surely that is not a controversial clam?

        • Swami says:

          Benwave,

          Sorry, but your claim actually is wrong.

          First, median wages have increased by forty to sixty percent. Studies which try to force the stagnation narrative do so by taking biased samples of the work force, using the wrong measure of inflation, excluding benefits taxes and transfers, and/or converting to family units and then not adjusting for changes in marital rates, retirement or family size. My links to the studies are at the top of the page, about the third comment from the top.

          I have read dozens of studies which track income by quintiles and they all show incomes, consumption and living standards are up.

          Even more importantly, we need to remember that individual families constantly move between quintiles or classes. So not only are the class averages going up, but over their lifestage’ most people move from the lower to higher quintiles before retiring and living off assets and thus going back to low income.

          What is bizarre to me is that something can be so conclusively proven and yet conventional wisdom is that stagnation “is not a controversial claim.”

    • anonymousskimmer says:

      I have a vague feeling that it is generally agreed that the last hundred or so years, give or take, have been an uncharacteristically wealthy and egalitarian era.

      I’m being US-centric, here. Literally the last hundred years, sure, but the baseline is the highly unequal roaring 20s. Here are some more analyses of the US GINI index since the founding:
      U.S. Income Inequality: It’s Worse Today Than It Was in 1774
      American Incomes 1774-1860 “In terms of inequality, our estimates suggest that American colonists had much more equal incomes than did households in England and Wales around 1774. Indeed, New England and the Middle Colonies appear to have been more egalitarian than anywhere else in the measureable world.”
      America Is the Richest, and Most Unequal, Country “the U.S., which also has the largest wealth inequality gap of 55 countries studied, according to the report.” “(OECD) examined income inequality, it found that the U.S. has the fourth highest income Gini coefficient—0.40—after Turkey, Mexico, and Chile.”

      • christhenottopher says:

        OK I’ve got a beef with that article. It says that income inequality was lower in the 13 colonies than today’s but first provides us with a gini coefficient measurement of .437 when in the modern US has a gini coefficient (as of 2013) of .41. Then when they want to compare the 1774 US to the present they throw up percent of income by income quintiles. OK but this disconnect seems a bit serious. Beyond that there is another potential explanation for increasing under-reporting of income by the lower end of the income distribution: the high marginal tax rates of combined progressive income taxes and means tested transfer programs. This has been increasing over time leading to a larger and larger disconnect in the US between inequality as measured by income and inequality as measured by consumption. The current system is highly complex with regards to who gets hit by what types of marginal tax rate changes, but particularly for the poor these rates can actually get over 100%. But of course that’s only for reported income. If you can dodge reporting (examples might include informal odd jobs with cash payments, under reporting cash tips, and illegal money making activities in cash, I’ve seen estimates that as much of 12% of GDP for the US is in the black/gray market economy), then your tax rate for that goes to zero. Given the relative lack of a progressive income tax and means tested income transfer programs at the time of American Revolution, I would be very surprised if income under reporting was higher back then.

        So is the current US more unequal than 1774? Well I don’t have comparable consumption data to work with so I can’t be sure, but that article sure failed to convince me.

  22. negative_utilitarian says:

    I don’t think it makes much sense to argue from history that automation doesn’t cause unemployment.

    Imagine you have a bucket with a hole in the bottom, suspended several feet above the ground. Water is constantly pouring into this bucket, but faster than it can drain out from the hole. As a result, the bucket is full, with extra pouring over the edges. Every minute, the rate with which water is being poured into the bucket decreases very slightly however.

    Mr. Strawman looks at this bucket, and posits a fundamental law, that the bucket will always remain full, no matter how much the output of the faucet drops. This theory would seem pretty solid for several hours, right up until the minute that the rate of water pouring in drops below the rate of water draining from the hole, causing the bucket to empty out.

    Unfortunately, the level of understanding we have of this phenomenon is not sufficient to allow naive empiricism to be of much use. Human history is filled with sudden reversals of laws that had held for the entirety of our history prior, due to changes that are often murky, subtle, or seemingly innocuous. The probability that the correct model of employment and automation is one that has sudden discontinuous shifts around the point where we are in some space of varying economic metrics, vs the probability that it isn’t, doesn’t seem especially easy to discern.

    The discussion that we need to be having at this point is “What model is most appropriate?”

    • herculesorion says:

      This is pretty much handled in the blog post, which suggests that what automation is doing is moving middle-income/skill/status jobs to lower-income/skill/status ones without changing the overall rate of employment or labor-force participation.

      • pontifex says:

        Scott does explicitly consider the idea that the “model” may abruptly change, though:

        They also say that robots will be able to do all human tasks, including novel-writing, science, and further AI research – sometime between 2050 and 2150. At that point, obviously, all bets are off, and we have a lot more than unemployment to worry about.

  23. The Big Red Scary says:

    Why should we simply write-off women’s increasing workforce participation as due to “unrelated cultural trends”?

    Any change in workforce participation must be at least partly due to economic factors. For one reason or another, there must be some incentive for women to work more outside the home than in the past, and some incentive for people to hire women more than in the past, otherwise neither would be happening on a large scale. Surely there is a feed-back loop between cultural trends and economic incentives.

    • Women have always been a somewhat sizable amount of the labor force, but the trend began to pick up in the 20th century due to WW1 and WW2 and men being off to war and also casualties, causing a temporary domestic labor shortage.

    • Tracy W says:

      some incentive for women to work more outside the home than in the past,

      Washing machines.

      • The Big Red Scary says:

        An example of technological employment!

        But that only partly explains the supply side. From the demand side, there is more to explain, given this observation under the graph disaggregating labor force participation for men and women:

        But throughout this time, men are leaving at almost the same rate, leaving only a moderate net participation gain… Around 2000, all the women who want to be in the workforce are there already, and the declining male trend takes over for a net decline.

        If most of those men leaving the labor force and being replaced by women actually wanted to leave, then employers would have to hire women to replace them, whether they liked it or not. Or perhaps the jobs they were doing simply ceased to exist and were replaced with other jobs, taken on by women entering the labor force. It would be very interesting to see the graph disaggregrated by industry. In cases where jobs continued to exist, could there sometimes be a demand side explanation that men were replaced with women? I’ve heard people say that male secretaries used to be fairly common until people realized they could pay women less to do the same work.

        I know it is not the point of this post, and surely economists have written many books on the topic, but the disaggregated graph was so striking to me that it just seems bizarre to pass it by so lightly.

    • Worley says:

      The “obvious” economic factors are:

      1) The remarkable automation of housework: the washing machine, dishwashers, vacuum cleaners, and the like.
      2) The movement of food-preparation work into factories and restaurants. (I have a copy of the The Joy of Cooking, and in the front are instructions for disassembling a freshly-butchered chicken.)
      3) The dramatic reduction in the amount of for-pay employment that is the exertion of brute force. (I am amused that when physicists analyze the operation of a steam engine, the output of mechanical energy is labeled “work”.)

      Item (3) is probably very important — one of the few ways men differ from women is in brute strength, and over the last 50 or so years, the economic value of exerting brute strength has gone from a substantial fraction of the economy to basically zero.

  24. sohois says:

    I believe it was Gwern who made the argument that even figures such as PAMLFPR were poor ways of trying to understand total employment, since such categorization is very much an invention of the modern era.

    As I understood their argument, even 100 years ago ideas such as retirement and compulsory education simply didn’t exist (note: am not historian, I don’t know when exactly compulsory schooling or retirement would have become widespread); anyone who wasn’t very wealthy would start working as soon as they were physically able, and continue until they were either dead or physically unable, more likely the former given health outcomes in those times. Thus, technological unemployment did occur following the industrial revolution, because it allowed for a huge number of former workers to stop working. However, economists did not recognize this as a decline in employment since new categories of education and retirement were introduced to absorb them.

    I can’t recall if Gwern provided figures to back this up, and I’ve not looked into it any more myself, but it certainly sounds a convincing argument and I wonder if anyone else would agree.

    • Tracy W says:

      However, economists did not recognize this as a decline in employment since new categories of education and retirement were introduced to absorb them.

      I think this muddles up demand and supply. Higher wages allowing a lot of workers to retire (or get more education) is not the same as there being a shortage of available work in the sense of unemployment. There’s plenty of jobs that could be done, even with modern automatation levels – e.g. greater healthcare, repairs on buildings. The issue is skills availability and willingness to pay, not a shortage of work per se.

  25. sideshowcrispin says:

    I’m sure you’ve covered this elsewhere, but what do you mean by “temporary bad period”? The experience of the US Rust Belt or the UK’s former coal mining areas suggest temporary can mean at least 40 years. Of course the world can change a lot in 40 years so quite possibly another overlapping “temporary bad period” has started by that point.

    In the language of economics, the system is permanently in a non-equilibrium state, so arguments based on equilibrium should be treated with caution.

    • Worley says:

      And I’ve noticed that in a lot of feel-good stories about “cities that came back” etc., there is always a period of 20 years or more. I suspect that most turnarounds involve generational turnover in the population.

  26. Luke Perrin says:

    93% of economists with an opinion on the issue agreed that the economic benefits of robots will be so great that they could be used to compensate the workers who were negatively effected. But in a survey I conducted in my imagination, 100% of people who have not been living in a cave the past two hundred years agreed that this will never happen in real life.

    Just increase incomes taxes and have a UBI.

    • Mark Atwood says:

      Just increase incomes taxes and have a UBI.

      I’ve seen the numbers for this, and it’s not pretty.

      I’d like to see you do them, because I doubt anyone else doing the math will persuade you.

      • Edward Scizorhands says:

        It’s why I say we should do wage subsidy instead.

        Paying three bucks for up to 40 hours of labor makes a lot more jobs worthwhile and significantly increases the job-supply for low-skilled workers.

      • Luke Perrin says:

        I’ve seen the numbers for this, and it’s not pretty.

        Do you mean numbers based on the current economy? Because those are irrelevant here. I’m talking about a future where (by definition) productivity has been vastly increased by technology. It seems to me that if

        93% of economists with an opinion on the issue agreed that the economic benefits of robots will be so great that they could be used to compensate the workers who were negatively effected.

        then there must (if you agree with the economists) be some level of income tax that will do that. Otherwise what does “could be used to compensate” even mean?

  27. Regarding the horse example, the US horse population is 9 million. The demand for horses never went away despite automation and has remained steady. If one tried to extrapolate the early 20th century horse population trend to the future, there should presently be no horses alive.

    Although the labor force participation rate is falling, the total labor force size has not (although it has flattened in recent years).

    Rather than the total elimination of jobs, imho a more probable concern is the bifurcation of the labor force by IQ, with those with an IQ below a certain threshold barely getting by, stuck with low-paying jobs that barely pay ends meet. Due to economic and technological factors, this IQ threshold will likely gradually keep rising. More then ever, IQ differences are tantamount to socioeconomic differences.

    This is why the decline of manufacturing jobs has not hurt the overall labor market size, because such semi-skilled, medium-IQ jobs are being replaced by service sector jobs. However, the pay for these jobs is not as good.

    Often what happens is the new technology will coexist with the older one, as the bank teller example shows. Other times the implementation is so slow that it is not a concern. I remember in 2007 McDonald’s, Jack in the Box, and other fast-food chains began to install self-order machines, but the adoption was really slow, the machines were often broken, and the machines didn’t replace the role of the cooks and cashiers. People needed help using the machines, necessitating staff. Same for those food store self-checkout machines. Rather than the machines eliminating the cashiers, they coexist. Some people like machines, but others don’t. Machines break and someone has to supervise the machines in case someone needs help or tries to fool the machine by imputing the incorrect weight or item.

    Overall, I am not too worried but I can see some cause for concern. Regarding post-scarcity and UBI, entitlement spending has been on an unstoppable rise for the past 50 years. We’re already headed in that direction. SNAP, medicaid, disability, and section 8 housing is pretty close to a UBI and post-scarcity. The future is one where the government assumes a greater paternal role in helping those who are unable to keep up with an increasingly competitive economy.

    • arlie says:

      This may be a digression, but the idea that the division is/will be based on IQ is a vast oversimplification, as any member of a disadvantaged group can easily intuit. It’s also a very comforting illusion for people who see themselves as having high IQ.

      Substitute “education” for “IQ”, and you may be a getting a bit closer. Especially if “education” explicitly includes credentials.

      But in the real world, who gets jobs easily – and who gets considered for “good” jobs – varied based on all kinds of things. Some of them correlate with intelligence, in part because a smart person may research them and manipulate those under their control. Likewise, how well they do once hired – and hence whether they stay in the job, or get a decent reference – also varies based on all kinds of things.

      Someone earlier in this thread posted about people getting fired for fucking in meat lockers. I very much doubt the individuals involved did this primarily because of limitations in general intelligence, let alone IQ. Poor impulse control seems far more likely. Or massive dysfunction in understanding of cultural norms of what is and is not appropriate … beyond the point of intelligence and well into pathology.

      Other people have trouble for more sympathetic reasons. They “don’t look the part” (a Hispanic engineer of my acquaintance reports regularly being mistaken for a janitor). They are introverted, or shy, or bad at blowing their own horn. They are stuck living somewhere with few opportunities, due to family responsibilities. Family poverty prevented acquiring educational credentials. They have an irrelevant disability that makes them look bad. (One acquaintance walks with a cane, and has bad interview results even for desk jobs – but only since the injury; she thinks that visible aging also makes her less likely to be hired.) Net result – a warm bodies, low skill, low pay job, or nothing.

    • herculesorion says:

      ” If one tried to extrapolate the early 20th century horse population trend to the future, there should presently be no horses alive. ”

      Compared to the number of horses there used to be, there aren’t.

      “Same for those food store self-checkout machines. Rather than the machines eliminating the cashiers, they coexist.”

      Hah. In California that’s because the union got laws passed saying that certain categories of products could not be purchased at self-checkout machines, so you must keep checkout workers if you want to sell those products.

      “Machines break and someone has to supervise the machines in case someone needs help or tries to fool the machine by imputing the incorrect weight or item.”

      That’s one person to ride herd on a dozen machines, though, not twelve checkout-stand workers and twelve more grocery-baggers.

    • Doesntliketocomment says:

      The horse number was specifically looking at horses in England. The first number I could find for that puts the number of horses in the United Kingdom at 796,000, so a drop to roughly a quarter of the maximum population. Looking for US numbers, I found one for a 1900 population of 21.5 million, and a 1920 peak population of 25.2 million, here: USA Horse and Mule Population, 1900-1960 So the current US horse population is around 35% of its maximum.

      The important parts I think you’re overlooking here is that it’s very easy to control horse populations. In the same time frame, the US population has increased 300%. Had horses matched us one-for-one in that regard, we would have another 66 million horses to employ.

  28. ricraz says:

    There’s no conceivable way the androids would “augment” human labor and there’s no conceivable way the displaced humans could go into another industry.

    This isn’t necessarily true. Significant parts of many jobs are driven by social interactions and status considerations. The job of doorman at fancy hotels may be both the most easily automated job in existence, and also the last one to actually be automated. In this particular scenario, it depends on how people relate to androids. If society has the general idea that interacting with androids is low-status and not as meaningful as human interactions, then there would be an explosion in customer-facing jobs, as it becomes the easiest way to differentiate your product. Also, everything produced by androids would be so cheap that people wouldn’t need to work very long hours in those jobs to earn a living.

    • herculesorion says:

      Haw. So what you’re saying is, there’ll be plenty of jobs for poor people…dancing at the whim of rich people who want to conspicuously spend money.

      • A Definite Beta Guy says:

        As Tyler Cowen would say, those new service sector jobs

      • ricraz says:

        Also things like making handcrafted goods, being artists and musicians, running events like meetups, being sports coaches or players, and generally playing a partly-social role in subcultures, like running hobbyist shops. I’ve written about this more here (scroll down to the section headed “The continued economic relevance of humans”). This is pretty speculative of course, but as long as humans place value on interacting with other humans (even when interacting with AI is indistinguishable), I think there will be some sort of economic role for humans.

        • AG says:

          John Barnes’ novel “Losers in Space” posits a post-scarcity society where everyone except for celebrities is restricted to a baseline level of luxury. Work like scientific research is therefore only done by those with the passion for it, but people can vote with their money on who is most entertaining them. So for the people who want a lifestyle beyond the baseline, it’s livestreamed status-signalling all the way down.

        • Kevin C. says:

          And what about those of us who are totally terrible at “playing a partly-social role”?

        • herculesorion says:

          Please to be reading entire thread before the responding.

  29. melboiko says:

    This was a pleasant lunch-hour read, thanks.

    I’m not trying to make you do my homework. I’m just saying that I’d really enjoy an up-to-date review, similar in length and spirit to this one, on issues related to the ongoing ecological problems; not just accelerating climate change but ocean acidification, topsoil erosion, permanent biodiversity erasure, the the sixth mass extinction etc.

  30. Nick says:

    Few comments unrelated to each other

    1) The numbers you are using for labor participation and disabilities are not age-adjusted. There is demographics working within 25 – 54 aged males and working age population is getting older, see: Median age of the labor force, by sex, race and ethnicity. I’m not sure if the story changes, but it should be taken into account.

    2) You seem show some puzzlement over the issue that can be explained by comparative advantage and declining labor share. Even when automation replaces jobs and does work better than men, comparative advantage virtually
    guarantees that everybody can find something to do if we ignore temporary friction in the job market. The only question is how compensation changes Only those workers who reach the minimum wage lower limit must exit the labor market.

    If everything else staying same, automation creating unemployment is the wrong question. We should look at the labor share. Automation versus labor is the story of capital replacing labor as factor of production, not automation removing working hours. Labour share has been in steady decline in OECD countries since early 70’s. (It’s probably coincidence that it starts dropping just when microprocessor has been invented.)

  31. xentronium says:

    Third, it is very hard to find a temporal correlation between the apparent effects of automation on manufacturing, and the rise in PAMLFPR. PAMLFPR has been rising very steadily since the 1960s and doesn’t seem to have noticed the manufacturing collapse

    Should that read PAMLFNPR instead of PAMLFPR?

  32. David MacKinnon says:

    Pardon my ignorance, you mention underemployment by way of skills, but not hours. At least in Australia, the casualization and rise in part time jobs in the labour market has been a periodic theme. And while that’s great as a choice, it isn’t always the case. In a (pt) job, looking for another job doesn’t make a dent in these stats.

    There’s also a huge difference between “making the ends meet..just” in an old style job for life vs “scraping by on most weeks, extra few dollars occasionally” as an uber driver who may be replaced tomorrow by a self driving. I would imagine there are are fairly large behavioural differences in those situations, even if, on paper they are both employed and, on average, earn the same every year.

    The data on this is so coarse and lacking useful nuance from study I’ve seen.

  33. micje says:

    > The argument in favor: look, imagine there’s a perfect android that can do everything humans do (including management) only better. And suppose it costs $10 to buy and $1/hour to operate.

    In a post scarcity world, we won’t need to worry about unemployment – that’s the goal! Presumably robots can create themselves, so what’s scarce in this example, if not labor?

    • Doug S. says:

      Raw materials and time – the things robots need to make stuff, including other robots.

      • Controls Freak says:

        At least in the case of items necessary for making robots, those things were capped at a $10 initial investment and <$1/hour maintenance costs, by supposition. If you're talking about other goods, then I’d be curious to know why you think that a scenario which assumes an across-the-board reduction in marginal cost of production will result in an increase in price.

  34. Denver says:

    And suppose it costs $10 to buy and $1/hour to operate. Surely every business owner would just buy those androids, and then all humans who wanted to earn more than $1/hour would be totally out of luck. There’s no conceivable way the androids would “augment” human labor and there’s no conceivable way the displaced humans could go into another industry. So at some point we’ve got to start getting technological unemployment.

    The logic here doesn’t follow. Yes, you could use said android to make a profit on any work over $1. However, assuming there is a scarce supply of androids, and a near infinite set of problems for humans to conquer, these androids would more likely be put to the highest productivity tasks they have a comparative advantage in, which doesn’t necessarily imply leaving only those jobs less than $1 to humans (unless there are a near infinite supply of androids which breaks the scarcity assumption).

    I’ve already tackled a similar problem here.

    • sharper13 says:

      Exactly.

      In regards to “a near infinite set of problems for humans to conquer”, this is a point typically underappreciated. There isn’t really a practical limit to human desires. If we all owned our own solar system of nanobots doing anything we wanted, people would be busy figuring out how to extend that to their own galaxy.

      There is no shortage of work to be done in the foreseeable future. All cheap robots+AI gets us (if they haven’t taken over from us) is the ability to accomplish so much more that our current wealth levels will seem like dark ages poverty in comparison.

  35. Matthias says:

    But maybe this is better viewed as a race between two competing forces; generational churn producing students with the right set of skills, and technology making new skills obsolete. I don’t know why this should have increased recently, but it seems like – at least for the middle class – this is a race they are now losing.

    This is exactly how Goldin & Katz phrase it:

    U.S. educational and occupational wage differentials were exceptionally high at the dawn of the twentieth century and then decreased in several stages over the next eight decades. But starting in the early 1980s the labor market premium to skill rose sharply and by 2005 the college wage premium was back at its 1915 level. The twentieth century contains two inequality tales: one declining and one rising. We use a supply-demand-institutions framework to understand the factors that produced these changes from 1890 to 2005. We find that strong secular growth in the relative demand for more educated workers combined with fluctuations in the growth of relative skill supplies go far to explain the long-run evolution of U.S. educational wage differentials. An increase in the rate of growth of the relative supply of skills associated with the high school movement starting around 1910 played a key role in narrowing educational wage differentials from 1915 to 1980. The slowdown in the growth of the relative supply of college workers starting around 1980 was a major reason for the surge in the college wage premium from 1980 to 2005. Institutional factors were important at various junctures, especially during the 1940s and the late 1970s.

  36. zima says:

    Great article, but I believe any discussion of technology and the economy has to consider the effects on the consumer side to get a full picture, which are clearly and significantly positive. If we produce more widgets with fewer people, the price of widgets will fall until people can afford more and better widgets to a degree that could compensate for lower wages. We do see historically low inflation and historically high improvements in product quality and variety (which means the real quality-and-variety-adjusted inflation rate could be even lower), and manufacturing (which has seen the most job loss) is also the industry that has seen the most improvement in products. Looking at the economist surveys and history, it seems virtually unanimous that the benefits to people as consumers of automation significantly outweigh the costs to people as workers.

    Right now, things are tough for ex-manufacturing workers because their industry has been automated but others haven’t, so those workers aren’t earning high wages but still have to pay high prices for products produced by less automated industries like education and health. If AI really gets good enough to automate everything, my prediction with 80% confidence will be that the benefits to people as consumers far outweigh the costs to people as workers (this also distinguishes people from horses because horses don’t control their economic consumption choices).

  37. mscantrell says:

    Maybe Donald Trump is going to look behind the White House couch one day, and find a large portion of the male population of the Southeast under the cushion.

    This was completely hilarious, and I literally laughed out loud. But I know you care about the quality of your writing, Scott, so… if it matters to you, your metaphor is jumbled up. You either look behind the couch, or you check the cushions.

  38. dfboyd says:

    In section I, you show a graph of PAMLFPR which peaks in the 60s and shrinks to 88% by 2014. “This is the decline people are worrying about when they talk about technological unemployment”. In section II you show the graph again and say “PAMLFPR has been going down consistently since 1960 or so”.

    Then you say “Third, it is very hard to find a temporal correlation between the apparent effects of automation on manufacturing, and the rise in PAMLFPR. PAMLFPR has been rising very steadily since the 1960s and doesn’t seem to have noticed the manufacturing collapse.”

    WTF?

  39. vV_Vv says:

    I am not entirely sure how differences in cognitive ability fit in here. My guess is to a first approximation they don’t – if standard economic theory is correct, it should be possible to create middle-paying jobs that use the full potential of people with any amount of cognitive ability, taking advantage of various human cognitive skills that are difficult to automate. Although some naive takes like “everyone should just become a programmer” fail to understand this, I don’t think the entire argument is based on misunderstanding of this point, or that it forms a particularly strong counterargument.

    For most of recorded history, almost every human was a subsistence farmer, likely under-using their cognitive abilities, yet no middle-paying jobs materialized to use their full potential. Then, during the first century or so of the industrial revolution, most people in industrialized countries were assembly line workers, jobs even less cognitively demanding than subsistence farming, and still no middle-paying jobs appeared until most of that low-skilled factory work could be automated.

    There is nothing in standard economic theory that predicts that jobs that allow average people to use their full cognitive ability should exist and pay middle-class wages. In fact, there is nothing that predicts that a middle-class should exist.

    • albatross11 says:

      If automation/AI causes a floor on employable IQ, my intuition is that it will be a relatively short period in human history terms (decades?) between when the village idiot is unemployable and when Albert Einstein is unemployable.

      • vV_Vv says:

        In the long run possibly, but in the short run automation seems to threat average-IQ jobs more than both high-IQ and low-IQ jobs, resulting in job polarization.

    • m.alex.matt says:

      For most of recorded history, almost every human was a subsistence farmer, likely under-using their cognitive abilities, yet no middle-paying jobs materialized to use their full potential. Then, during the first century or so of the industrial revolution, most people in industrialized countries were assembly line workers, jobs even less cognitively demanding than subsistence farming, and still no middle-paying jobs appeared until most of that low-skilled factory work could be automated.

      There is nothing in standard economic theory that predicts that jobs that allow average people to use their full cognitive ability should exist and pay middle-class wages. In fact, there is nothing that predicts that a middle-class should exist.

      That’s because, for most of recorded history, most people lived under social conditions that were far too politically insecure to foster the development of a market economy. The moment you get conditions moderately close to safe, you start developing prosperous peasants, middling craftsmen, and increasingly wealthy traders in burgeoning small towns and everything goes from there.

      Seriously, be careful about importing social conceptions of class into economic analysis. Standard economic theory certainly DOES predict increasing incomes in a lot of different conditions, which manifests in reality (amongst other ways) as a growing middle class.

      • vV_Vv says:

        That’s because, for most of recorded history, most people lived under social conditions that were far too politically insecure to foster the development of a market economy. The moment you get conditions moderately close to safe, you start developing prosperous peasants, middling craftsmen, and increasingly wealthy traders in burgeoning small towns and everything goes from there.

        I think you are inverting cause and effect. In order to have large numbers of craftsmen, traders, professional soldiers, guards, etc., anybody whose full time job is not to grow food, then those whose job is to grow food must produce enough surplus to feed them.

        If almost everybody is a subsistence farmer and eats almost all the food that they produce themselves, then there is very little stuff to trade, no matter what the political conditions are.

        • m.alex.matt says:

          ‘Able to produce only enough foodstuffs for subsistence’ is not some iron clad law of the material conditions of subsistence farmers, it’s a result of the dynamics of the social and economic conditions they live under. Being able to produce enough of of a surplus to rise above subsistence conditions is exactly a result of stable political conditions that incentivize investing in future productivity instead of in current security (building a wall versus building a mill).

          Given a few years and peace it’s surprisingly easy for peasants to become small scale commercial farmers. Doesn’t matter how they do it, they’ll find a way.

          • vV_Vv says:

            Given a few years and peace it’s surprisingly easy for peasants to become small scale commercial farmers. Doesn’t matter how they do it, they’ll find a way.

            So why didn’t it happen for most of recorded history? It’s not like there was constant war at every place and every time, and even when there was war, its impact on rural peasants (most of the population) was quite limited, unlike 19th and 20th century total wars that entailed war economy, mass conscription and depopulation attacks on civilians, and yet happened during the time when the middle class rose.

          • John Schilling says:

            It’s not like there was constant war at every place and every time,

            No, but wars every generation or two seems to have been a common pattern. And the existence of a large community of prosperous small-scale commercial farmers seems like it might incentivize neighboring communities to do some large-scale raiding. Unless you’ve got institutions that can stop that.

            and even when there was war, its impact on rural peasants (most of the population) was quite limited, unlike 19th and 20th century total wars that entailed war economy, mass conscription and depopulation attacks on civilians,

            I think you have a profound misunderstanding of how war was waged in the preindustrial era. When e.g. English Yeomanry and French Chivalry marched off to settle their disputes in one glorious day (er…) at Agincourt, what do you think they ate?

            The answer involved something euphemistically called “foraging”, and it was very much not good for the peasants. And that’s before we get into sieges, which were as common and as decisive as battles if not more so, and the scorched earth techniques deliberately practiced to prevent the enemy from foraging. and ravaging as a deliberate policy of military-economic coercion.

            Preindustrial war was very much not a polite game that the warrior caste played among itself and with little consequence to the peasants except that the tax collector might fly a different flag next year. Preindustrial war tended to reduce the peasantry to the bare minimum, in numbers and/or surplus wealth, necessary to feed what is left of the victorious army. That’s not good for economic development.

          • m.alex.matt says:

            So why didn’t it happen for most of recorded history?

            It did. All the freaking time. Where do you think cities came from? Literally any time a large enough area experienced stable enough socio-political conditions for long enough you got the development of economic specialization and trade that led to increasing wealth all around. For a few generations in a row sometimes, even! In some bright, infrequent flourishings you might even get centuries of the stuff, where civilization would grow into what people would look back on as a golden age in hindsight.

            I think the point I’m trying to make is that you have no idea what economic history actually looked like. You’ve got this weird, Hollywood vision of medieval Europe that you’re generalizing into all of world history. That isn’t how things have ever worked, except in the most awful, horrid conditions where public security absolutely collapses and people are so desperate to survive until tomorrow they stop worrying about a decade from now.

            What you might call a middle class (in analogy to what we call middle class these days, not the historical usage referring to a non-noble bourgeoisie) is really freaking common, historically. What is special today is the sheer scale of wealth there is to go around, not the general idea of wide participation in commerce.

  40. Yosarian2 says:

    PAMLFPR has been going down consistently since 1960 or so (has it accelerated recently? probably just the recession). And the number of US manufacturing jobs only started to really go down around 2000.

    Looking at your numbers we may have 2 different things happening here. Maybe from 1960-2000, the decrease in male employment was caused by the increase in female employment; and from 2000-today, female employment has been more flat, and the decrease in male employment has been caused by automation? Does that correlate better to the numbers?

    • Brad says:

      That’s exactly what I was thinking of at that part. What may have changed in 2000 is that we ran out of untapped potential women workers.

  41. greghb says:

    Sure this post is long, but is it long enough?

    If you’re taking requests for follow-ups / came across materials you’d recommend: this post leaves me wondering about earlier instances of rapid technological advances that dramatically changed the make-up of the labor force. Are there case studies about the Luddites (or, I don’t know, farmers in 1910), from which we might judge how much these groups look like modern men in WV?

  42. maintain says:

    You said MFLPR once when you meant to say MLFPR.

  43. alwhite says:

    I wonder if human consumption has been adequately accounted for and if we’ve reached a maximum of how much we can consume? When we freed people from farming, there was a lot of other stuff that we wanted and could turn are focus towards, thus employing people. Now we have Netflix. When all of our excess time has an outlet, what more could we need that would employ people? Parallel to the idea that today’s robots are betters than yesterday’s Rolodex; today’s desires are better met than yesterday’s desires and there’s just not enough demand to drive us beyond what technology has already done.

    I’m not entirely sure how this ties in, but I think it’s a related idea. The story of the bank teller gives an important view of how jobs are changing. They’re becoming more relationship and service oriented. We no longer need people to count money, we need to people to build relationships. We no longer need someone screw toys together, we need a conversationalist while they cut our hair.

    An economy is more of a abstract idea than a hard physical fact. That means we can theoretically create whatever kind of economy we need. So we could shift the economy to have mostly human relations type employment while the machines do all the physical stuff, but right now, Americans don’t value those kinds of roles. We don’t want to pay for services, we want to pay for things.

    Also, why should I pay to watch your highly trained skill in ballet when I have Netflix?

  44. benquo says:

    Urbanization causes unemployment because urbanization is the shift from work to jobs, and has been happening over the right timeframes: https://en.wikipedia.org/wiki/Urbanization_in_the_United_States#Historical_statistics

    To be a bit more specific – there’s no “unemployment” on the family homestead, there’s just either the opportunity to take a bit of a break, or the opportunity to make some long-run improvements. Unemployment can only happen in urbanized environments in which labor is a commodity with a market price.

    Any economist worth their salt will tell you that cities, not robots, drive productivity gains. Robots are just one particularly dramatic example of the sorts of specialized actuators cities can make if you let them do their thing for a few millennia.

    It seems like we’ve been experiencing substantial technological – i.e. urbanization-related – unemployment since the 1800s. Socialism (in particular, French-style limits on working hours and layoffs) and Keynesianism (clever gimmicks to trick people into spending more money, creating more “demand” for labor) have been the two credible contenders for solutions (the third, central planning by an authoritarian state, has been discredited, not expressing an opinion on how fair this was). The US’s taste has been strongly towards Keynesianism, so we’ve been masking the effects of technological unemployment for at least a century.

    • Wrong Species says:

      So do all inventions that originate outside of cities not count as improving productivity?

      • benquo says:

        They count. Why would you think otherwise?

        The urbanization hypothesis is not a claim that cities exist in a vacuum.

        • Wrong Species says:

          Wow, I was not expecting that answer. Either I’m somehow misunderstanding you or you don’t know what productivity means. If Robinson Crusoe crashes on an island and builds himself some kind of animal trap, he has increased his productivity. This isn’t debatable. You’re just flat out wrong, in the same way that calling Socrates a chicken is flat out wrong.

          • benquo says:

            Fuck it, I’m not trying to comment here anymore, if this is the kind of response I get.

          • Iain says:

            You are talking past each other.

            It looks like benquo is talking about “urbanization” in the sense of “a move away from subsistence farming towards paid labour”. In that sense, the division is not about whether an inventor lives inside of a city; it’s about whether the inventor is a subsistence farmer, or is working within a framework of paid labour.

            I don’t think benquo is arguing that Robinson Crusoe (a subsistence farmer) can’t increase his productivity. I think the argument is that lone castaways on desert islands are going to see less productivity growth than societies in which people have paid jobs.

            @benquo: When somebody says “Either I’m somehow misunderstanding you, or [you are wrong]”, you will get further by trying to clarify the confusion than by storming off in a huff. If you are willing to try the former, I encourage you to stick around.

          • Wrong Species says:

            @benquo

            In my morning haze, I both misread your comment and turned up the snark too far so I apologize. Don’t let me being an asshole taint your experience here.

          • benquo says:

            Apology excepted, thanks 🙂 I’ll give it another try, if you think there’s any substantive disagreement left to talk about here.

            Iain, to be fair, I am making a slightly more specific claim: that this has something to do with measured rates of urbanization, i.e. people moving towards population centers. But this is entirely compatible with a lot of productivity enhancements originating in the rural periphery. Indeed, for some agricultural innovation, this is almost surely true sometimes. This process, however, tends to be very, very slow (a slight improvement in agricultural productivity doesn’t necessarily let a subsistence farmer outbreed their neighbors by very much), unless there is also lateral information transfer, i.e., unless there’s a mechanism by which the innovation spreads to other people. Cities somehow seem to serve that function, and in principle could do so even if all the actual inventions had to come from outside.

            Occasionally, such innovations can spread through an alternative process; military conquest is the obvious one, for innovations with clear military applications such as the iron chariot, horses big enough for mounted cavalry, and formation and drill.

          • Doesntliketocomment says:

            @Wrong Species
            What you are using as “productivity growth” is just expected use of manpower. We assume that a hunter/gatherer/farmer will use the tools at their disposal to produce as much as they can. At a certain point however, they will not be able to produce more with only their own efforts, no matter how clever. Ten thousand years of subsistence farming did not produce super-farmers who could out compete industrial methods with clever devices made from sticks. The “Urbanization” benquo is talking about is related to the division of labor- As a farmer, it’s impossible for me to dig and smelt my own iron without jeopardizing my ability to farm, but if I could get an iron plow it would increase my productivity enormously. Urban centers are important because they provide the networks that enable things like iron and coal and crops to be exchanged in a way that increases productivity. The more urbanity you have, the more complicated the exchanges you can produce, the higher overall productivity.

            @Iain
            I see what you are saying, but I think trying to continue conversation in the face of rudeness only normalizes rudeness.

          • benquo says:

            Doesntliketocomment, thanks for saying that RE Iain. I was looking for a way to point out that evenhandedness is obnoxious, but was struggling to figure out how to do so in a way that didn’t accidentally throw shade on Wrong Species, whose apology I’m actually satisfied with. I hope this apophasis works.

            ETA: I meant “apology accepted” above. Sorry!

    • benquo says:

      One correction: urbanization drives changes in productivity gains. There had to be some initial productivity surplus to make cities in the first place.

  45. baconbits9 says:

    I have longer thoughts overall, but one main point that deserves mentioning is that a lot of categorization is semi arbitrary and misleading (warning: Historically inaccurate example follows)

    Example: X years ago agriculture was Y% of the economy where Y was very high. Despite years of productivity increases in agriculture Y remained high, how can this be? Well first roughly 100% of employment was in agriculture, small groups of people farmed small plots of land. A new invention increases productivity, what do the farmers do with the surplus? Some of it gets sent to a different group of farmers in exchange for some of their surplus. How to get it there? A couple of the farmers take it in a cart pulled by oxen. That is a new job, the oxen drivers are probably the ones caring for the oxen on the farm as well, and maybe mending the carts in addition to helping out with the harvest at peak times. A farm that has a peak need of 20 or 30 pair of hands during harvest finds all kinds of work for them to do, repairing the roads, cutting fire wood, grinding grain, or any number of things.

    Eventually productivity growth is high enough to create a whole new industry. The farmers that chopped wood during down periods eventually give way to loggers who chop wood all the time, the farmers who would drive the oxen and trade eventually split off fully, and then as productivity rises further the drivers split into drivers and merchants. At some point it is clear- the merchant never spends a day of his life on a farm, and he trades many things that are not produced on the farm so he can’t be an agricultural worker, but it is non obvious when the split happens.

    This goes on an on. The grain that was once ground/processed on the farm is now shipped to the newfangled mill a few miles away along the river. The Miller needs help loading the mill, filling bags, shipping them etc, but of course there is down time as well, so the Miller’s helpers pick up other skills, one of them learns to shod the horses coming in when there is a need, eventually traffic is so high that he no longer has time to help load the mill and is a full-time blacksmith. At roughly the same time it has become economically viable for the mill to invent, or copy some new invention, a labor saving device, those that haven’t grown enough eventually peter out and close down while those that do offer cheaper services that attract more grain shipments that require additional blacksmiths, vets, and hospitality workers in the growing town.

    This is Arnold Kling’s Patterns of Sustainable Specialization and Trade. The invention of the tractor which displaces farm hands also simultaneously creates demand for people to load and ship the greater amounts of grain, more millers, more salesmen, more bakers, etc, etc, etc.

  46. onyomi says:

    Not saying I necessarily think this is what is going on, but one simple possible explanation for why technological unemployment could happen now when it never happened much in the past could be quite simply the greatly accelerated pace of change.

    For most of history, technological change was very, very slow. The past few hundred years we’ve moved increasingly to a place where each new generation has to learn to function in a world different from the one their parents grew up in. We could now be moving to a world where each generation has to learn to function in multiple worlds over the course of a lifetimes, which may stretch the limits of human adaptability.

    Couple this with an educational signalling game keeping people in “training” for skills largely inapplicable to any job market, not uncommonly to the age of 25 now, and you have a pretty narrow window to master and apply the skills needed in the new economy before those skills become a lot less valuable and you run out of the time/energy/money/mental flexibility to learn a whole new paradigm.

    Though I definitely think overregulation, bailouts, monetary policy, etc. can stymie the necessary adjustments, arguably states have already been pushed by technology in a liberal direction by sheer necessity relative to e.g. guilds, or even old-fashioned unions, for that matter.

    • Edward Scizorhands says:

      The time frame from the invention of the cotton gin to humans actually being displaced out of doing that work was, what, 50 years?

      The time from the first self-driving car being good enough to navigate the roads to drivers being removed will be way shorter than that. I don’t know when it will start, but when it does, it will finish very quickly.

      • JonathanD says:

        Pedantry and somewhat beside your point, but you’re misunderstanding what the cotton gin does.

        When you pick cotton, it’s filled with seeds, which, pre-gin, had to be picked out by hand. This was slow and very labor intensive, so cotton was a cloth not much in use. The gin made the process easy, allowing cotton to be processed much more cheaply. The gin’s adoption was rapid and widespread, and it led to a massive increase in the use of cotton (and, incidentally, slaves). It was invented in 1794 and essentially no one in 1844 was picking out seeds by hand, that would have been insane.

        You’re thinking of the technology for picking cotton. I don’t know anything particularly about that, but I can tell you that it showed up sometime after WWII. (Source: my grandparents and older cousins grew up picking cotton, before that process was automated.)

  47. mario rossi says:

    Isn’t running all these analysis at the national level a mistake? The world economy is more integrated than it ever was. Containers (among many other innovations) lowered transport costs. We had a massive increase in the world’s middle class population in the last 20 years or so. Why are we complaing about the middle class disappearing when the opposite has happened at a global level?

    We used to have a massively bimodal distribution in world income, mostly driven by technological and political barriers. The world economy integrated and the world income distribution moved to the middle. Looking at the effect of this change on US or Western population is bound to be distorting. We cannot extrapolate the effect of integrating developing countries in the western economy over the next 20-30 years, there is no equivalant mass of population to integrate.

    I understand why the national level is interesting, but I think in this case it’s not really helpful.

  48. Lasagna says:

    Another great article – thanks Scott! I had two comments:

    This is potentially consistent with a story where the jobs that have been easiest to automate are middle-class-ish. Some jobs require extremely basic human talents that machines can’t yet match – like a delivery person’s ability to climb stairs. Others require extremely arcane human talents likewise beyond machine abilities – like a scientist discovering new theories of physics. The stuff in between – proofreading, translating, records-keeping, metalworking, truck driving, welding – is more in danger.

    This is one of those statements that sounds plausible, and you’ve certainly backed it up, but just doesn’t match anything I’ve seen in real life. Yeah, yeah, “anecdote”, whatever: at some point, the conclusions reached by analyzing data need to match experienced reality.

    Let’s take me (a one-time high-powered lawyer) and my cousin (low-skilled worker in a factory creating medical devices in New England).

    I’ve watched the legal industry’s upheaval over the last twenty years, and can say with certainty that AI has helped wreck the path that lawyers from top schools formerly took to wealth and security after graduation. Basically: corporations used to pay ungodly sums for dozens of ivy league-educated attorneys to review documents for 12-14 hours a day, sometimes for years straight; now the review is largely done by computers, overseen by a mid-level associate or two and MAYBE a small team of contract attorneys. This has made “getting wealthy through white-shoe law firms” much more difficult than it used to be – there was always competition for top firm spots, but now the supply of highly-educated lawyers is far outstripping the demand for them. Yes, there are a lot of reasons why the legal industry collapsed in the last ten years. This is one of them, and is the one that’s getting worse, not better, as time passes. So “lawyerin'” is at least one example of automation wrecking a highly paid industry. I was able to shift to an equally highly paid position when I saw the writing on the wall; I know plenty of people who weren’t.

    My cousin works in a factory. When he started there twenty, twenty-five years ago, they employed a thousand workers or so. Now they employ about fifty, partly because of offshoring, partly because of automation. He survived by being the perfect worker – never takes sick days, never takes vacation. He will be laid off eventually, and his skills are not transferable.

    Actually, now that I’m writing this, I realize that he fits the “middle class” example of a job being destroyed, so he fits with your thesis. I should delete it, but it kind of segues into my next point:

    On the other hand, there are signs of technological underemployment – robots taking middle-skill jobs and then pushing people into other jobs. Although some people will be “pushed” into higher-skill jobs, many will be pushed into lower-skill jobs.

    Genuine question: how do you get “pushed” into a higher paying job? What’s the trick? My cousin has four kids, and would love to know.

    • baconbits9 says:

      Are there more or fewer people employed in the various forms of ‘lawyerin’ than 20 years ago?

      • Lasagna says:

        Good question, and I can’t answer it. But I can say this:

        1. Not all forms of lawyerin’ are equal, so once you include “various” types of lawyers instead of the narrow subset of highly successful ones I was talking about you get into a lot of other issues. A lot of people these days are graduating with JDs and finding that the jobs available to them do not make enough money to pay back their loans (the number of people here, in New York, working as interns for DAs and doing full-time work for sometimes zero salary is infuriating). What I was discussing was the drop in opportunity for people who previously made lots of money – the high earners in the industry. But there is also a massive drop in wages for lawyers lower down on the ladder who will still come up as “employed” in the statistics.

        2. I assume you’re asking (in part) if using AI to review documents instead of highly compensated attorneys is leading to more people in the AI document review world to have a job than highly paid attorneys to lose theirs. I can only give my instinct: highly doubtful. An increasing amount of this work is being outsourced to India, so shit jobs in India are replacing highly paid jobs in the US. I don’t think that’s a good thing; you form your own opinion. But the stuff not being outsourced is not resulting in more employment in other industries here – it really is AI running the show. You buy the software, a paralegal scans the documents, and one or two attorneys instead of fifty or a hundred oversees the results. This is absurdly simplified, obviously.

        Closing thought: if you’re going to propose that the loss in attorneys might be offset by an increase in people making, upkeeping and selling the AI software, again, my experience is that you’re wrong. These industries haven’t gone on a hiring surge; the software is basically sold by the same companies that used to do document management for attorneys, but were being driven out of business (and cutting staff) because of automation and AI.

  49. Jiro says:

    If relationship banking is profitable when tellers are freed up to do it, then it would also be profitable if tellers are not freed up to do it and you hire new tellers to do it instead, which means they would have been hired anyway. And having tellers move to relationship banking still is technological unemployment if the relationship bankers would have been hired anyway.

    It’s like hiring a painter then saying “I decided I don’t need any paint” and offering to have the painter be my landscaper instead. If I was going to hire a landscaper anyway, that’s still down from two jobs to one job. Saying that not doing painting “freed up the painter for another job” is just an accounting trick that allows you to not count the painter losing the job by having the landscaper lose a job instead.

    Also, this post shows that Scott needs to use brevity. Most of the graphs in this serve the function of only 1) making it unreadable because the graphs include more information than is actually relevant, and 2) suggesting “I’m right because I have all these graphs and you don’t”. There should not be more than 2 or 3 graphs here; anything else needs to be a link only. I gave up halfway through because of the graphs.

    • baconbits9 says:

      If relationship banking is profitable when tellers are freed up to do it, then it would also be profitable if tellers are not freed up to do it and you hire new tellers to do it instead.

      This isn’t automatically true.

      • Jiro says:

        I can’t think of a reasonable scenario where it isn’t true, except a narrow one related to not having to train the bankers as much.

        • JayT says:

          Say a teller makes $10/hour and “relationship banking” brings in $6/hour worth of value. At the same time, say an ATM removes half of the teller’s workload.

          In this case the bank wouldn’t have hired someone to do the relationship banking since the cost of the employee would be higher than the value of the job. However, since the ATM freed up half the teller’s time, they were able to use the teller to do the job that wouldn’t have otherwise made sense to do.

          • Iain says:

            “Freed up half the teller’s time” is another way of saying “made it possible to hire half as many tellers”.

            Paying somebody $10/hour to produce $6/hour of value is not going to work out well for you.

          • A Definite Beta Guy says:

            That teller isn’t producing $6/hour, they are producing $11/hour, because the ATM removed half the workload.

            That means you will then hire more tellers, since they are now producing $11/hour, not $10/hour.

          • Jiro says:

            If the teller starts out producing $10/hour worth of value, and the ATM reduces the teller’s workload, the teller is now producing $20/hour and the bank doesn’t need as many tellers, so half the tellers get fired.

            If instead of firing a teller, you put the teller on relationship banking, the teller is doing work with $20/hour productivity part of the time and work with $6/hour productivity part of the time. That is no different than firing half the tellers, having the remaining tellers do the $20 productivity work all the time, and hiring people to do the $6 productivity work all the time.

          • JayT says:

            Beta Guy gets at what my point was. If it made sense to hire a teller at $10 an hour, but then half their job was eliminated by the ATM it would then make sense to get rid of half your tellers. However, If the teller can do both their original job (which is now half as time consuming) and the relationship banking job (which was only worth $6/hour because it also isn’t very time consuming) at the same time, then that teller is now doing a job that would make sense for the bank at $11/hour.

          • Iain says:

            No, that still doesn’t work.

            This story only works if there is some constraint preventing people from doing relationship banking all day, or if we somehow postulate that spending half your time working as a teller has no impact on your productivity as a relationship banker. If $6/hour relationship banking is sufficiently low-effort that your tellers can do it along with their old jobs, then you could instead hire full-effort relationship bankers who would presumably bring in $12/hour. And you could have done that without ATMs.

          • Randy M says:

            The reason they have the time to do the new $6 per hour worth of value job, the relationship banking, is because now much of what they had been doing is being done by the ATM. So they are providing half the value to the employer (approximately) that they used to, plus whatever value from the new activities.

            That doesn’t mean they can tack on the new stuff and be contributing (at either hours a day) $88 worth of value. It means that they will contribute 4×10 + 4×6, or 64$ worth of value, at an expense to the company (ignoring any non-monetary compensation and cost) of the original $80 per day.

            Now, in the real world, maybe they have to be on site for 8 hours to periodically double check the ATM totals, or have an extra pair of eyes to reduce risk or some other random reason. In that case, it makes sense to use up their time in any profitable way possible. But the bank will be looking to consolidate those positions as soon as it works out logistically.

            Most likely, if the banks really did retain the same number of emmployees, it is because something in the market changed making those new positions profitable, or “relationship banking” was a new innovation that would have paid off earlier if it were tried, or there was some regulation change as well.

          • A Definite Beta Guy says:

            They aren’t doing the relationship banking along with their old jobs, they are doing relationship banking alongside half of their old jobs.

            ATMs and Tellers aren’t substitutes. A Teller performs job functions A,B,C. The ATM performs functions A and B. This leaves the Tellers doing function C, and then the banks assign them do function D as well. Or all tellers might get fired and Function C gets absorbed into the Relationship Manager positions.

            Plus, we’re talking about value at margins here. If A,B,C,D are all complementary functions, then ATMs doing more A and B makes C and D a lot more valuable. It is the same way that the existence of a steel factory suddenly makes automobile manufacture more feasible.

          • baconbits9 says:

            A beta guy is on the right track.

            2 types of customers, people who want fast, efficient service and those that want personal banking. With tellers handling everyone they are stuck in the middle ground. Rush everyone and you alienate those that want the personal service, be to friendly and deliberate and the other group is unhappy. A teller who spends half an hour with a customer who wants it will lift their head to see a line filled with grumpy people who just wanted to deposit a check and go.

            Install an ATM and the teller can assume that everyone who walks in the door wants the personal touch. The people who want personal banking are more likely to be happy waiting 15 mins or so as long as they get their half hour in when its their time. It seems like it should cut teller numbers way down but once they are free to spend longer periods of time with customers that want it (and thus will pay for it) you barely see any drop at all.

        • baconbits9 says:

          The first is that the automated banking is part of making personalized banking profitable. Automated banking would provide a natural funnel for dividing the customers into people who just need a quick transaction and people that want a personalized one. This way your well trained and highly social tellers aren’t spending large chunks of the day performing dull, low value and repetitive tasks (which also might end up sapping some of their qualities for the customers that need them).

          • sharper13 says:

            Exactly, you have to consider both supply and demand and the market clearing price.

            At a lower price point, normally customers will demand more, but sellers will be willing to supply less, because their marginal cost makes them unable to profit on the higher number of customers at that lower price point.

            However, if checking, savings and credit card accounts require less labor to service (because of automation in the form of ATMs and other computers), then they will cost less in terms of time and money for people to purchase. People who may have had a single checking account and savings account before may get one at multiple banks, plus a credit card and a line of credit against their house.

            So at the new, lower cost, the price point for that same service is lower, which means more customers will want more of it (to a certain extent). The higher demand at the lower price point (without less profits) means the supplier will supply more of it, hiring a corresponding higher number of employees to perform the non-automated portions of the work.

            When banking products which would not have been profitable to offer before (due to the marginal cost being higher than enough customers want to pay) begin to cost less, then they can be offered by the “tellers” at a profit, when they wouldn’t have existed before.

          • anonymousskimmer says:

            Every time I go into the bank it’s to make a quick transaction that can’t be done at the ATM. Usually this involves asking for money in amounts other than 20s (or 5s, when available).

            I’m sure some of the other people are just those who never adapted to ATMs, are paranoid that the machine will screw up their deposit, or etc…. For whom the phrase “relationship banking” makes sense.

          • baconbits9 says:

            @ anonymousskimmer

            The difference is still between all customers having to go inside the bank and a smaller subset going in. The proportion of people who want personalized banking among those standing inside the bank is bound to be much higher after the advent of ATMs.

        • Doesntliketocomment says:

          The problem isn’t with the tellers, it’s with the bank itself. The bank itself is a physical building and an organization of finite complexity. At a certain point, you can’t increase the number of tellers without having a physically larger and more complex bank, and these items don’t necessarily scale linearly. If you remember banking pre-atm, recall how crowded it was, with long lines despite literal walls of tellers. I am sure you did have “relationship bankers”, but they had to be shoehorned in. Now with the need for tellers cut down, and lines evaporating due to people banking off hours, you have plenty of floorspace for lower-margin activities. I would guess this is much of what produced the explosion of rural banks, as with ATMs handling the gruntwork, banks could operate with smaller footprints and staff.

    • JulieK says:

      If relationship banking is profitable when tellers are freed up to do it, then it would also be profitable if tellers are not freed up to do it and you hire new tellers to do it instead, which means they would have been hired anyway.

      In the first scenario, the ratio of bank employees to bank customers stay the same (but the customers get a better level of service). In the second, the customers have to support twice as many employees.

    • A Definite Beta Guy says:

      This is narrowly accurate, but not a correct characterization:

      Bessen observes that two forces worked in opposite directions. First, by reducing the cost of operating a bank branch, ATMs indirectly increased the demand for tellers: the number of tellers per branch fell by more than a third between 1988 and 2004, but the number of urban bank branches (also encouraged by a wave of bank deregulation allowing more branches) rose by more than 40 percent. Second, as the routine cash-handling tasks of bank tellers receded, information technology also enabled a broader range of bank personnel to become involved in “relationship banking.”

      You are complaining about the second. The important point in the second is that increasing technology not ONLY provided ATMs, it also provided a whole bunch of other things that made relationship banking possible. This made tellers useful, just a different sort of useful.

      What a teller does changes as more technology changes what job functions are possible and what job functions are automated.

  50. P. George Stewart says:

    Maybe the difference from past Luddism, and why it was always wrong in the past, is that markets always had headroom to make big innovations and there were always new markets to conquer in the geographical sense.

    But now the world seems to be saturated, and genuine innovation seems to be stagnant. For a while, there was still some possibility of horizontal progression (“create demand”, hypnotize people into consumerist zombies updating their toys every 5 minutes), but even that has limits, especially if people are so bound to the bankster/govt. debt-based Ponzi scheme that they can’t afford the toys.

    I can see the possibility of a positive outcome though: an economy of two halves, one which is automated (sophisticated phones as cheap as matches eventually), the other which remains human, where you have people on UBI able to afford the cheap basics, individually moseying around in their own good time to entrepreneuring themselves into personally fulfilling vocations, the products of which sell at more of a premium. The super rich still get their dosh from the vast, automated industries, but the mass makes a decent enough extra living (above their UBI) on the side, from human services, man-made things, bespoke Maker stuff, etc.

  51. Cerastes says:

    I think a big problem with answering this is that human data gets parsed down into categories and subcategories and sub-subcategories, but automation is simply presented a vast, amorphous, universal force with no such subtlety. In reality, it seems to me that there are several axes on which automation varies on, which have different impacts on the labor market. Actuator/power technologies determine how much force/work/power is available (or power/$), energy storage technology determines ability and duration of systems to operate “untethered” (in the literal sense). Sensor & control technology determines whether devices are “clockwork” or can respond to changing situations, and software/processing/AI determines how well they can integrate sensors, run image-recognition, learn to cope with new environments, etc. Fundamental advances is any of these areas would be expected to have different effects on different jobs, and different jobs would only become “obsolete” when different combinations of one or more categories of automation make advances. Horses got shut out just because of advances in actuation/power (steam and internal combustion engines), humans doing calculations (the original “computers”) were rendered obsolete by computers. But to replace a truck driver, we need advances is both sensation/control and software/processing/AI, and to replace a construction worker, you’d need advances in all categories (modern actuators still *suck* compared to biological muscle, and both fat and glycogen have energy densities an order of magnitude higher than the best batteries). So when you just lump everything together as “automation”, you just look at huge trends and miss that the internal combustion engine affected a different part of the workforce than the transistor did.

    And maybe that’s “why things would be different now”? In the past, advances in these areas were occurring slowly and happened at different times, allowing people to shift from job to job as needed, but now, as the changes come more rapidly in all areas, people won’t be able to adjust?

  52. paulsutter says:

    Stagnant productivity growth may be a misleading measure of the impact of technology. Manufacturing productivity is measured in dollars[2], and prices have been going down on manufactured goods[1] (likely due to competition from China)

    [1] https://ourworldindata.org/grapher/price-changes-in-consumer-goods-and-services-in-the-usa-1997-2017
    [2] https://www.bls.gov/lpc/faqs.htm#P03

  53. bkennedy99 says:

    I think there is something to the cognitive ability story. Tools offer a tradeoff of physical labor for cognitive ability (to learn to operate the tool). More productive tools require more cognitive ability. Over time, the cognitive requirements for participating in the modern economy ratchet upward, which begins to leave more and more people behind. To quote Warren Buffet, “Someone in America who has a 90-point IQ is qualified for many fewer jobs today than he was 100 years ago.”

  54. Erusian says:

    The argument in favor: look, imagine there’s a perfect android that can do everything humans do (including management) only better. And suppose it costs $10 to buy and $1/hour to operate. Surely every business owner would just buy those androids, and then all humans who wanted to earn more than $1/hour would be totally out of luck.

    No such machine will be built for a very simple reason. There’s overwhelming competition from the human sector. Imagine not that you’re in some mythical future but in the present. You’re Bill Gates and you’re looking for projects to fund. Someone says, “Give us a billion dollars of your money and we’ll make a machine that does everything a human does. Maybe. You know, if we do it right.” Bill Gates, not being an idiot, will answer, “Don’t humans cost $7.25 an hour?” You, thinking you’re very smart, will say “Yes, but these robots will cost $1.00 an hour.” To which Bill Gates replies, “So you’re saying there’s a net savings of $6.25 an hour. Maybe. If you succeed. Or I could go out and buy 160 million hours of the proven, already existing, millennia old solution. I could pay every Walmart worker minimum wage for a fiscal quarter. This seems like a solution in search of a problem.”

    (And even if Bill Gates did decide to push that boundary, the androids themselves would necessarily be scarce. Their labor, like anyone’s labor, would be bid for. Just because they only need to consume $1.00 an hour to survive doesn’t mean that’s what they’ll charge. Under the current economic scheme, they will be robo-capitalists (or owned by capitalists) who will charge as much as they can get for their labor.)

    Nor is there a finite number of tasks that need to be done. If you free up more labor, then workers will do other things. There can be frictional unemployment as the economy changes (people’s towns die out and they don’t want to move, people trained for a decade as a Copyist Monk and now that damn Gutenberg is doing it cheaply, people want new jobs but need to break in). This does not mean there will be unemployment forever. People will transition, or they will grow up seeing Joe Coalminer going broke and decide to do something else. Not just coding and other upper class professions.

    Maybe they go out west and plant an (infamously hard to automate) peach farm. Maybe they start making wooden furniture by hand and sell it as a luxury good. Maybe they take a degree in running the machines that put Joe out of work. Maybe they put up a little theater in rural Ohio and perform Shakespeare. Maybe they start building bigger and bigger houses so everyone shudders to remember when everyone had only (shock!) three rooms. As you say, these sort of jobs are more immune to the super-android scenario. If the robots are really better than humans at everything, why would waste their time swinging hammers when they could be doing something like curing cancer? They’re also all fairly middle class. (None of the jobs listed here have an average salary of less than $42k, Pew’s cutoff for middle class.)

    Also, keep in mind, if super-androids do everything better and cheaper, everyone is getting sold better goods more cheaply. This means, even if their income decline, they become collectively wealthier. If wages are cut in half but prices fall by 75%, everyone is twice as wealthy as they were. If robots reduce all labor costs to $1.00 per hour, then you might be left only making money delivering pizza, but that $8.00 an hour might buy you four televisions.

    (I won’t even get into how little sense a post-scarcity economy makes.)

    Grace et al survey top AI researchers on when AI might be able to replace humans in various things

    Top experts confirm their field is indeed valuable and about to Take Over the World. In other news, preacher claim God exists and the rapture is imminent. More at 11.

    Seriously, AI researchers have been saying this is just a few decades away since the ’50s at least. It’s not clear why we should believe them now.

    • Wrong Species says:

      Taking your Bill Gates argument to its natural conclusion, no corporation would ever invest in technology enhancing improvements because it’s trading a sure thing for a risk.

      • Erusian says:

        I appreciate the argument ad asburdum, but it misses the key point. The point is not uncertainty, it’s the financial calculation.

        If you said, “Hey Bill, give us a billion dollars and we’ll build a program that writes better software more quickly than the best coders in the world,” then Bill would say, “Those seventy thousand people with an average salary north of six figures? That I spend billions of dollars paying each year? I could give you five billion and, if it works, still save billions!” And you get funding.

        This is how automation decisions usually work. The famous Spinning Jenny cost less than the difference in wages between a skilled and unskilled weaver, etc. It’s why the minimum wage hike led to a stock price bump in automation companies.

        • Doesntliketocomment says:

          Firstly, Bill Gates doesn’t need 160 million hours of minimum wage labor, he needs a $10,000 minimum-wage-bot he can sell.

          Secondly, as you are considering the automation of industry, I don’t know if you’ve accounted for the industry of automation. What if our hypothetical billionaire made his money by developing ways to automate jobs? Now that he’s developed cheap automation for one industry, he has to move on to another if he wants to keep his business. If you think about it this way, what you have is a bunch of Google-type engineers trying to think of a way to automate hairdressers out of a job so they can get paid for the quarter.

        • Wrong Species says:

          Even without minimum wage increases, McDonald’s might decide to automate some positions if they can’t enough people willing to work for low enough wages. If your choice was between being unemployed or taking a job working $0.50 an hour, what would you choose?

          • Erusian says:

            Firstly, Bill Gates doesn’t need 160 million hours of minimum wage labor, he needs a $10,000 minimum-wage-bot he can sell.

            But the people buying minimum-wage-bot need minimum wage labor. The hypothetical startup is thus competing with minimum wage laborers. A one time $10,000 robot is indeed cheaper than a minimum wage worker, cheap enough to make it viable if it does exactly as much work. But it gets outcompeted by a $1,000 dollar solution that requires a $7.25 an hour minimum wage worker to operate and allows one worker to do the work of two. Why do complex, expensive AI research to completely replace humans? You can build a human-used solution more cheaply.

            (You also run into the issue that, if they require maintenance, minimum-wage-bot is inflexible, whereas you can fire and rehire workers. Which means you would probably see robo-temp agencies.)

            Secondly, as you are considering the automation of industry, I don’t know if you’ve accounted for the industry of automation. What if our hypothetical billionaire made his money by developing ways to automate jobs? Now that he’s developed cheap automation for one industry, he has to move on to another if he wants to keep his business. If you think about it this way, what you have is a bunch of Google-type engineers trying to think of a way to automate hairdressers out of a job so they can get paid for the quarter.

            I’m not sure I understand. Automation is only incentivized where it is profitable. If the automation is not more profitable for the consuming firm than their current labor intensive solutions, then this hypothetical automation firm would not get many customers. If you’re a hairdresser and your staff costs you $100,000 a year, and AutoCorp comes to you with a machine that costs more than that, then you’ll tell them to take a hike. And AutoCorp loses the money they spent developing the Hairdresser 5000.

            Even without minimum wage increases, McDonald’s might decide to automate some positions if they can’t enough people willing to work for low enough wages. If your choice was between being unemployed or taking a job working $0.50 an hour, what would you choose?

            McDonalds will choose to automate if automation is cheaper than labor. They won’t automate if labor is cheaper because they’re a company and will not generally take the more expensive option for the same result.

            But labor is relatively cheap. And where an incentive exists to automate, there’s little reason to expend the extra money to completely replace the human. Rather, automation usually allows one person to do more and more. What’s more likely to happen is that you will see one person running an entirely McDonalds making pretty good money, but McDonalds is paying less in total.

          • Doesntliketocomment says:

            What I am talking about is capital. Correct me if I’m wrong, but you seem to saying that nobody will invest the capital to develop a machine to replace low-cost labor while the labor is itself available. What I am saying is that the capital may already be invested, rather than being a new venture the company that would do this might already exist, and are simply looking for new ways to profit off their own human and technological resources. Each human job is not sui genris, and the lessons learned in one can be applied to another. The fact that people have poured capital into developing robots to replace pizza delivery drivers and hotel concierges shows that they are not afraid to aim low.

            Also, I think you are wrong about the $1000 tool vs the $10,000 replacement. Consider that in your example, all you’ve done is changed the price of the robot from $10,000 to $19,000, and the depreciation is still far less than the price of one minimum wage worker. Worried about staffing fluctuations? Robots can be rented, resold, or packed in a shipping container and shipped where they are needed.

            I am also unsure of what you are trying to say with the McDonald’s example. If one person can run a McDonald’s, then you’ve eliminated most of the jobs involved, especially since the one job remaining is most likely a higher-skill position.

          • Erusian says:

            What I am talking about is capital. Correct me if I’m wrong, but you seem to saying that nobody will invest the capital to develop a machine to replace low-cost labor while the labor is itself available.

            No, what I’m saying is that people will only invest capital if investing capital in automation is cheaper than the alternatives. And that ‘hire one of seven billion existing units’ is going to be cheaper than completely eliminating human jobs for a long, long time. Maybe forever.

            What I am saying is that the capital may already be invested, rather than being a new venture the company that would do this might already exist, and are simply looking for new ways to profit off their own human and technological resources.

            And what I am saying is that it’s irrelevant. If it does not make microeconomic sense for the companies they sell to, those companies will not buy it. Whether it’s developed in house or out of house, that’s the same. Even if some genius develops a perfect human AI, it needs to be economically viable. Otherwise it ends up the equivalent of the aeololipile.

            Each human job is not sui genris, and the lessons learned in one can be applied to another. The fact that people have poured capital into developing robots to replace pizza delivery drivers and hotel concierges shows that they are not afraid to aim low.

            My argument is not that the job has to be high or sophisticated, my point is that the economic calculus has to make sense. And that it needs to make sense to completely replace the humans altogether, otherwise it’s just a very sophisticated tool and should operate as other tools have historically.

            The future is not technological unemployment. The future is a series of increasingly technologically sophisticated tools freeing people up for more and more specialized, productive jobs. At least until society collapses.

            My point, in summary, is that there is a yawning chasm between ‘smaller groups of humans using tools to do more’ and ‘there’s nothing for humans to do’. And that it’s unlikely to be economically viable to cross that chasm. So long as we don’t, technological unemployment is frictional as people move to an industry, retrain, or their kids move away from dying communities. These people need to be supported and helped but it’s not a world ending, no one will work proposition.

            Also, I think you are wrong about the $1000 tool vs the $10,000 replacement. Consider that in your example, all you’ve done is changed the price of the robot from $10,000 to $19,000, and the depreciation is still far less than the price of one minimum wage worker.

            And the minimum wage is $15,000 a year, plus they’re eliminating another person’s job by doing twice as much. My numbers were chosen to demonstrate this point: If hiring a human (or hiring a human plus less expensive technology) is cheaper, people will do that. And it’s hard to see how it couldn’t be cheaper.

            Let’s imagine two pieces of software that… I don’t know, makes paperclips. Software A completely replaces the human and has a mind just as sophisticated, capable of making decisions just as well if not better. Software B takes a human’s input and makes paperclips exactly as it’s told to. Software B is almost certainly cheaper because it doesn’t need to simulate the mind of a paperclip worker.

            Worried about staffing fluctuations? Robots can be rented, resold, or packed in a shipping container and shipped where they are needed.

            However, once you buy the robot, you’re incentivized to use as much of their labor as possible. You’ve effectively bought all of the robots labor, whereas with an employee you are only buying what you need. This means there’s a huge issue if your demand fluctuates, even just throughout the day. You also have to pay the full upkeep no matter what. So whether this is beneficial is dependent on several factors.

            I am also unsure of what you are trying to say with the McDonald’s example. If one person can run a McDonald’s, then you’ve eliminated most of the jobs involved, especially since the one job remaining is most likely a higher-skill position.

            Somewhat higher skill, maybe, but only in the sense hotel concierge is ‘higher skill’ today because it requires knowing computers. Even if it’s a complicated machine, it’s not apparent to me that the fast food equivalent of, for example, operating a crane requires a college degree. I would not call the techno-machining specialists working in steel plants in Indiana upper class.

            The average number of employees at a fast food establishment is 15. Let’s say you have a team of three at the Auto-McDonalds rotating in and out. So you’ve eliminated twelve jobs. This would mean that Auto-McDonalds would be cheaper and it becomes economically viable to have more Auto-McDonalds, which means there would be some commensurate increase in the number of restaurants. Plus people can get jobs do something else.

            In fact, they can do something less economically productive and still have the same standard of living because Auto-McDonalds (and auto-Everything Else) is cheaper.

    • Cliff says:

      Sorry to make such a low quality response, but your argument is obviously absurd and wrong, and based on made up premises and assumptions. You randomly posit a $1 billion development cost. What if the development cost was $5? Then you posit minimum wage businesses, even though the quote specifically says the robot can do ANYTHING a human can.

      • Erusian says:

        We currently spend 12 billion a year on AI research and have yet to make anything nearly as sophisticated as the human brain. The one billion is an arbitrary number. It’s also absurdly low. You think someone will develop a human level AI for less than it costs to buy a pizza? $5 is literally less than Gates would have to pay someone to go and install the technology from a flash drive.

        I posited a minimum wage business because if the minimum wage jobs survive, some jobs have survived. In fact, I specifically mentioned that even if you had a robot that could do all jobs, unless there were infinite robots, humans would still do things that the robots didn’t have time to.

        I look forward to your higher quality rebuttal.

        • Cliff says:

          Yes but 12 billion has been spent which has gotten us much closer to these robots and it has been worth it, even though no such robots have yet been produced. Much of the work towards these robots will be a byproduct of researching other useful improvements.

          You used minimum wage businesses not to explain why some jobs would survive but to explain why such robots would never be invented. Obviously if the robots were invented to take over much more highly compensated work, then the game is over because the robots have been made, and once made there is no reason they should not also take hte minimum wage jobs.

          • Erusian says:

            Yes but 12 billion has been spent which has gotten us much closer to these robots and it has been worth it, even though no such robots have yet been produced.

            I’m not arguing the money is ill-spent. I’m pointing out that 1 billion dollars to make a human level AI is absurdly cheap, and so a startup that needs one billion to do it is a generous assumption.

            Much of the work towards these robots will be a byproduct of researching other useful improvements.

            Such as? What useful improvements lead to replicating human style intelligence?

            You used minimum wage businesses not to explain why some jobs would survive but to explain why such robots would never be invented.

            Because if humans are cheaper than robots, then they won’t be developed. Minimum wage businesses are as cheap as humans can legally be right now.

            A world where society-wide technological unemployment is a thing and a world where humans can contribute economically are mutually exclusive. If humans can contribute economically, then absent some other factor they will still have jobs.

            Obviously if the robots were invented to take over much more highly compensated work, then the game is over because the robots have been made, and once made there is no reason they should not also take hte minimum wage jobs.

            I suggest you read David Ricardo. Or perhaps someone more modern. It’s accepted economic orthodoxy, empirically proven, that even if you are worse at literally everything people still benefit from economically engaging with you.

  55. Disillusioned9 says:

    Excellent article!

    I have two comments: first, the automation of manufacturing may now have meaningfully affected the PAMLFPR nationwide, but surely the localized effects may tell a different story? The economies of manufacturing heavy States like Ohio are in crisis, which would strongly suggest that either PALFPR or underemployment would go up in these areas. Increases in non-manufacturing jobs elsewhere in the nation could smooth those numbers in the US and hide regional inequities.

    Second, maybe “this time it’s really different” not because automation or technological progress is unique, but the pace and magnitude of it are. In the not so distant future, the robots will render a lot of routine occupations obsolete, triggering both declines in PALFPR and increasing underemployment. Just as importantly, technology is developing at such a breakneck pace, in part due to advances in computing power under Moore’s Law, that our window to adapt without massive pain is incredibly short (it will be painful even in the best case scenario). And I think this time, it may well be beyond our fiscal and especially cultural will.

  56. Freddie deBoer says:

    Great post. I’m always a little bugged by people who say “if robots start doing X, the people laid off will get jobs fixing the robots.” But if maintaining the robots required the same number of people at the same wage, then you wouldn’t build the robots at all; there would be no economic motivation to do so.

    • albatross11 says:

      Yeah, I think the real argument is more like the one with immigration: Immigrants or robots or outsourcing will take some existing jobs away, but then the whole society will be wealthier, and that extra wealth will create new jobs somewhere else.

      I suspect that this is broadly true in the long run, but:

      a. There’s likely to be a lot of short-term pain involved.

      b. There’s a lot of path dependency in a society, so we could end up with a new equilibrium situation that was a lot worse than the current one, after everything’s come back into balance and all those displaced $20/hour unionized steelworkers are reemployed as $subsistence+epsilon/hour call-center employees.

      c. There may be things that just permanently exclude a lot of the kind of people who were formerly employed–effective floors on intelligence/education for employment, a society where everything’s on your permanent record so someone who sold drugs at 19 is unemployable at 30, underclass dysfunction so horrible that it leaves most of its kids unable to hold down jobs anywhere, etc.

    • Cliff says:

      “But if maintaining the robots required the same number of people at the same wage, then you wouldn’t build the robots at all; there would be no economic motivation to do so.”

      Only if the robots were not more productive than the people. If you can produce more for the same cost then you will generally want to do that.

  57. Jack Lecter says:

    that ridiculous Coal Miner To Coder scho

    You’ve expressed this sentiment before. Is coding really that hard? I’m honestly asking.

    I’m inclined to defer to your expertise on this (you actually know people who code) but it seems surprising. I would have thought there were at least some jobs that didn’t require high IQ- checking old code for mistakes, that sort of thing. And I would have thought at least some Coal Miners wouldbe high IQ, even if it’s not most of them.

    So what am I missing?

    • Freddie deBoer says:

      1. The kind of abstract thinking required to code is that hard, yes. Depending on how pessimistic you want to be, something like a third to a half of American high school students can’t do algebra to their state-mandated requirements. Which are pretty low.

      2. Contrary to what people think, computer science is not an everlasting jobs fountain.

      • Lasagna says:

        Good to see you posting again, Freddie!

        I’d also add that there’s the extra step rarely discussed: our freshly-scrubbed 40+ year-old newly minted coders also need to leave Kentucky and West Virginia to go code somewhere else. That means not only finding a job that can pay to haul their families far away from home (assuming that our middle-aged coder isn’t, for example, divorced, and would lose the opportunity to raise their kids if they move), but it means that it’s possible to interview for and get a low-level coding job in California from your Kentucky suburb. And that everything else about your life is worth leaving in order to get an entry-level coding position. And so on.

        • Freddie deBoer says:

          Precisely so.

        • Edward Scizorhands says:

          People in Boston and San Fran will object, but there are lots of coding places elsewhere in the country. You could even work remotely from Kentucky.

          I still think Freddie and Scott are right in that you can’t really made everyone a coder, but if it works they don’t necessarily have to leave town.

          • Mark V Anderson says:

            I think it is very difficult to get your first job in any new profession from home. In probably any profession, and definitely in a skilled profession like coding, one really needs to be closely supervised for the first few years before one is up to speed. I suppose in theory this could be done remotely if that was the only option, but I don’t think demand for these jobs is so high that employers would be willing to do this. Anyone tell if I am wrong.

            I am sure there are actual jobs in coding firms in Kentucky, but somehow I doubt they are very plentiful. Maybe the top coder in a miner to coder school will get one of these jobs, but I can’t imagine this working for most.

        • Cliff says:

          Precisely not so. Lots of coding is outsourced to India. Someone can’t work remotely from Kentucky? This is not a personal service job.

    • Iain says:

      As an aside: “checking old code for mistakes” is one of the most grueling and difficult parts of coding. It’s hard enough to build your own mental model for how to solve a problem; it’s even worse trying to retroactively deduce the mental model of the person who wrote the original code, and figure out where it went wrong. (This is true even when that person is your past self.) Even better, code that you have a reason to go back and look at it is disproportionately likely to be poorly designed and documented — if it was better code, it probably wouldn’t have so many bugs in it.

    • ilikekittycat says:

      As someone who has TA’ed for computer science at a state university where the first year is “Pre Computer Science” because so many people change major, it’s not even guaranteed that you can teach coding to the subsection of the population already willingly seeking that out as a life/career choice.

  58. herculesorion says:

    “The best-paying jobs – managers, professionals, and the like – are doing fine. The lowest-paying jobs, like personal care and food, are also doing fine. It’s the middle-paying jobs that are in trouble. ”

    So it’s back to the basic theme of the modern economy: Fuck The Middle. The wealthy (big companies, people with lots of money, workers with high-demand skills and experience) can absorb the impacts just fine. The poor (tiny single-employee craft businesses, people with no money, workers who’ve been low-skill their while lives) either have enough of a safety net that they can bounce, are small enough to not be noticed, or aren’t significantly affected by the changes.

    The middle gets hosed. Mid-size companies are at a critical point where just a small increase in costs can leave them without enough income to keep operating, and they don’t have the capital to stay in the game until they can grow out of the problem. Mid-wealth people–own one house but the mortgage is eating all their income so they can’t save, don’t have significant investments–can’t absorb sudden unexpected costs (medical, home repair, sudden auto replacement or family emergency). Mid-skill people–tradesmen and clerks, who’ve learned a skill on the job–get boned by technological or economic changes that leave no demand for people with their skill.

    The middle was always where economics went to die, and the modern economy has been set up to grow the middle faster and faster.

    • Swami says:

      I disagree Hercules,

      People are not assigned permanent classes. Data shows that people tend to start in the bottom quintile especially while going to school or just starting a job; move up to middle tiers as they graduate and/or get experience and/or get married (gaining second income); then they move up to higher quintiles as they hit peak experience years before moving back down to the lowest quintile when retiring and living off assets.

      Certainly there are a small share of families who spend their entire lives around one or two narrow quintiles, but this is the exception and not the rule. Broad statistical reviews completely miss the trend as experienced by most American families.

      The data clearly shows (I can link if desired) that the larger trend happening over past thirty years is the middle class is shrinking primarily by moving into higher tiers or retiring, the lower class has grown slightly, but not if you exclude thirty to forty million immigrants. And here much of the statistical increase has been caused by lower numbers of married families (and thus fewer jumps to middle income due to failure to get hitched).

  59. John Schilling says:

    If there used to be lots of work for horses, and there is still lots of work for average humans, do mind the huge gap between the cognitive abilities of the brightest horse and the average (or even two-sigma dull) human. There was, and presumably still is for the time being, a great deal of work that needs to be done in that gap.

    The improvement of automation has historically occurred within that gap. As such, the people whose particular jobs have been rendered obsolete have pretty much always been able to step up to “better” (more cognitively challenging, more productive, less wasteful of human potential) jobs. Meanwhile, the effective labor force increases to consist of humans + more and more machines, with humans capable of claiming all of the extra productivity. Everybody wins. But when the wave front from automation starts to cross the limits of average human capability, that’s going to be a qualitatively different scenario. So I’ve never held with the people who just say “automation has always made things better, why worry?”

    What is curious to me, and I hadn’t really noticed it until your deep dive, is that automation has started crossing that threshold in middle-class rather than lower-class jobs, which would suggest that the lower-class jobs are more intellectually challenging than the middle-class ones. And yet they pay less and carry lower status. My guess at what is happening is that the range of challenges faced by e.g. a janitor or a home health aid is, while not terribly deep in intellectual terms, sufficiently broad that no school or training program can substantially improve on the common sense that just about any human brings to the job, and no AI “expert system” can yet cover all the bases. So it’s a job that anyone can do (hence large supply relative to demand, low wages, low status), and so far no machine. But the intellectual challenges of middle-class jobs – and this is going to be particularly true in manufacturing – are deeper but narrower, such that a college or trade-school education offers a substantial advantage. But also such that the problem can be bounded in a way that lets us program and train an AI to cover all the relevant bases. It’s the jobs that a few narrowly trained people (hence limited supply, high wages/prestige) can do, that the machines can do first.

    Leading to technological underemployment as those people are pushed into the low-but-broad cognitive demand jobs. That could lead to the sort of class divide we haven’t seen since the industrial revolution if it persists, or towards mass technological unemployment if we soon get robots capable of doing those jobs as well.

    • Randy M says:

      automation has started crossing that threshold in middle-class rather than lower-class jobs, which would suggest that the lower-class jobs are more intellectually challenging than the middle-class ones.

      There’s definitely different kinds of intellectual challenge with humans and machines have different comparative advantage for at the moment.
      Think of an accountant versus a someone replenishing items at wal-mart. The Wal-mart employee probably won’t be able to fill out more complex tax forms, especially not reliably. The types of calculations and the details that need to be paid attention to are harder. Even though it has basically already been automated for most cases.
      Whereas noticing visually which items are missing in a display and pleasantly greeting customers as he refills them plays to a lot of innate human hardware and software.

      • JayT says:

        I think even more than that is the fact that there is more incentive to replace the middle income employee than the low income one. If creating technology for filling out a tax form and restocking a shelf were both to take similar amounts of effort, then you would obviously focus on the job that costs more to fill. Once you have most of those middle income jobs automated, then it would make sense to try and automate the low income jobs, but not before then.

        • Randy M says:

          Well, of course it is a combination of the cost and expected pay-off. But the cost of adapting a machine to a job consisting primarily of numbers has got to be less than adapting it to visual recognition, mobility, etc.
          And the reason low income jobs are low income is that they are better suited to what humans are innately (ie, without developing unique skills) good at.

        • anonymousskimmer says:

          Another reason is real estate.

          Automate a data crunching job and you can reduce your real estate costs while delivering a noticeably cheaper product to your clients (which will leave them still relatively happy even if the automation makes a mistake or two).

          Automate a store-shelf stocker and you still have the real estate cost of the store, and likely have to demo and reorganize the interior of the store to allow the automated solution to function (without inconveniencing shoppers) while still having ideal layout for product packing (how much product you can fit in x square feet) and shopper experience.

          This, itself, is an expensive task, with tons of room to screw up in such a way that your company permanently loses market share – especially in such a competitive industry as shopping.

          And all you really gain is the wages of a half dozen shelf stockers who can also do double duty helping shoppers and working the registers. This might allow you do decrease the cost of your products single digits – something a typical shopper may very well not notice.

    • herculesorion says:

      These kind of jobs aren’t hard to do. What they are is hard to plan. If you give me a person and guarantee that it’s always going to be the same size and weight of person, with the same sort of ass, and it’s always going to have the same consistency of poop in the same places, then sure you can build a bum-wiping robot. But there’s so many requirements there that a machine to handle every possible input would be incredibly expensive, even though actually wiping the bum is not a difficult task.

      It’s the same old story; you use a machine for something that’s easy to plan but hard to do well, and you use a human for something that’s hard to plan but easy to do well.

    • Andrew Hunter says:

      I make a year to sit in a room and write angry emails and the occasional kernel patch. I’ve been doing it my entire adult life and I’m pretty good at this; I deliver quite a lot of value to my employer and capture a small fraction for myself, on which I live like a king (sans courtiers, sadly.)

      I’m not sure I’d make it a year as a barista without getting fired. I have a temper enough to get noticed in a high octane development environment where mouthing off is most likely to get my boss to try to calm me down and suggest I’ll be more productive if I don’t; I’m pretty sure pulling that at Starbucks gets me canned. I am horrible at maintaining a schedule and being on time; ditto. I have some serious social anxiety problems; a job where I have to walk up to strangers and ask them for money sounds like a recipe for panic attacks (though who knows, exposure therapy might work? This has certainly gotten miles better than it was.) Oh, and as disability goes, I have iffy posture, flat feet, and some recurring back issues which mean standing up for more than an hour or so leaves me in quite a bit of pain.

      I think my vague point is: there are a lot of axes on which jobs can be hard for different populations. I’m not sure how to measure the number of jobs and the number of people qualified or unqualified based on any of:

      – g-loading
      – repetive strain
      – social interaction
      – civility
      – time orientation
      – status anxiety

      or half a dozen other axes. 🙁

      • Andrew Hunter says:

        A couple more semi-independent ways in which people can differ in ability:

        – outgoing social skills
        – (at least code switching to) high class signifiers (think: banker)
        – toleration of travel (and family circumstances that allow it)
        – coordination
        – physical strength
        – artistic talent (many subdivisions here)
        – submissiveness

        The list goes on and on. What’s most interesting to me is that for most–but not all–of these automation is actually reasonably equally risky, it just shows up in different ways. Computers aren’t great singers, but they mean every pub no longer needs to hire a top notch vocalist to have music playing. They’re not great conversationalists on phones, but automated menus do work. Roombas don’t care if you sneer at the help. Etc.

        So it’s much more a question of something else about whether a particular job that requires some subset of these things can be automated or not, and thus whether people with those skilsl but not others get wedged out.

        • RobJ says:

          This is part of why I question the conclusion that automation is likely causing underployment, but not unemployment. With so many dimensions to what makes a job automatable, it seems unlikely that it’s targeting the middle class so specifically. Not to mention that the middle class would uniformly move to upper or lower class jobs rather than out of the workforce.

        • herculesorion says:

          It’s also a question of whether customers are willing to accept reduced capability in return for reduced cost, and the answer really does seem to be “yes”.

          Albeit it’s often a perception of reduced capability, and the reduced “cost” is intangibles like time or personal hassle. As in, you were *never* going to get an extension on your driver’s license renewal, but at least with a computer menu there’s no illusion that you could talk the clerk into doing it…and you can be pretty sure the computer isn’t going to laugh to its buddies about how sad you looked.

  60. Swami says:

    My simple answer after following the topic for several years…

    The reason men are not working formally as much as they did is because they don’t have to. As we get wealthier as a society it easier for to get by without a job.
    1). You can live off your parents who are wealthier than prior generations
    2). You can live off girl friend(s) or spouse, who now work (not so much a few generations ago) or get public assistance.
    3). You can live off-the-books. Working for cash on the side, Craigslist, black market sales, etc.
    4). You can claim disability even if not disabled for all possible jobs
    5). You can go to school often using the GI bill or student loans. This may just be passing time.
    6). You can retire early and live off assets created earlier in your career (and remember median US wages are higher than any time ever, especially when corrected for taxes, transfers and PCE inflation)

    I certainly agree that technology and lower competing wages from abroad (and you forget 40 million immigrants) are reducing the wage rate-of-increase over prior generations. But this just changes the costs and benefit ratios. The rewards of working are lower for men and the opportunity to get by without working are better than ever. Mix the two trends and a subclass of men are learning how to get by without it showing up on the books as formal employment.

    With women we see a similar trend of having children outside of marriage. As men become increasingly optional due to female employment and public assistance, women are have been having more kids outside of marriage. And since there is no marriage contract, this feeds right back into the pattern of men not having to work to support their families. Single guys don’t have to work.

    I am not a person living in a post graduate academic or professional bubble. My family and friends are loaded with young, able bodied men living just as I described above. They shift between categories, sometimes living off the books, sometimes going to school, sometimes working a year or two, sometimes reselling deals on Craigslist, sometimes getting disability or veteran benefits, sometimes scrounging off women or parents.

    This is a thing.

    • Jaskologist says:

      Came here to say something close. A big factor missing from this post is the decline of marriage. Married men are the engine of the economy.

      • Wrong Species says:

        Another pertinent fact: immigrants have much higher participation rates than natives. That holds even when you adjust for class. There are jobs available if you really want them. I guess the question is whether the decline came more from cultural factors or economic ones.

    • 2181425 says:

      A related and interesting doc from MIT on this issue:

      The Emerging Gender Gap in Education and Labor Markets

      (warning: pdf)

    • albatross11 says:

      One question I have:

      There is a set of people, mostly men, who are unemployable because they have outstanding judgments against them which will be quickly garnished if they get an above-ground job. This includes a lot of people with child-support judgments, and probably other stuff I don’t know about. There is also a largish set of people, again mostly men, who have outstanding warrants/judgments for fairly minor criminal violations. (These are especially common in places where some local government is running their police department as a revenue-raising operation.) Again, these people don’t dare work above ground, for fear of either having their paychecks taken away or having themselves hauled off to jail[1].

      I have no idea how large a chunk of people that is, or how much impact they have on statistics. I’d expect them to be pretty hard to measure–certainly, if I were trying to avoid going to jail on an outstanding warrant, I’d never fill in any official forms about my location.

      Does anyone know whether that set of people is an important part of the story w.r.t. lots of working-aged males not having jobs?

      [1] I keep thinking US society is about one more generation from reinventing debt peonage and debtors’ prisons. And then thinking I may be optimistic.

      • Swami says:

        Great additions to the list albatross. The increases in incarceration and child support both work against finding a formal job. Better (in their views) to work in the trillion dollar plus informal/black market or not at all.

        I really think the globalism and technology angles are not explaining much of the underemployment issue (they may contribute though). It is better explained by changing marriage rates, higher prosperity, larger social welfare benefits (for men and their child bearing girlfriends), higher incarceration rates, changes in child support, and changes in cultural expectations.

        I think there is a reflexive tendency to attribute this underemployment to stagnation. But what if the real problem is a slow building disease of affluence? The US is currently wealthier than 99.999% of all societies ever. And by a significant, not subtle amount (current incomes are twenty to thirty times above the historic 2-3 dollar a day living standards. Maybe what we are seeing is the sociological effects of assortative matching and slowly changing institutions and cultural and mindsets.

    • Worley says:

      I suspect that Swami is onto a big aspect of PAMLFNP — as the economic pay for brute labor goes down, it’s no longer guaranteed that a man who is willing to work hard can earn substantially more money than almost any woman. If men are motivated to dedicated employment primarily because they can then support a wife, then men who are unable to earn enough to be a substantial economic benefit to a wife (vs. what she can earn herself) won’t be motivated to be employed.

  61. AnteriorMotive says:

    Scott’s missing an important effect, which is that the same rising productivity which drives down wages will help people buy more with their increasingly marginal jobs.

    The agricultural revolution pushed 98% of the population into marginal jobs (all of ours, when compared to the all-important task of feeding people). But it also made food so dirt-cheap that people with frivolous jobs like “bank teller” can buy more than they can eat.

    You can imagine a world in which all but one human skill are obsolete. It /still/ wouldn’t be the apocalypse, because no matter how unimportant it is, the robocapitalists can afford to spend millions on it.

    Hell, imagine a world in which /every skill/ is obsolete, and you have to undercut the labour androids. Yes, you have to work for 50c per hour, but the massive productivity gains in the rest of the economy mean that 50c per hour can buy you a life of luxury.

    (Zima makes a similar point above)

    • Rick Teller says:

      I’m amazed that of 140+ posts so far, you and Zima are the only ones that noticed what Bastiat called “that which is unseen,” in this case the impact on consumers of being able to get what they want cheaper.

      We see the worker getting laid off because robots or workers overseas can do their job cheaper. But why do businesses automate or switch to overseas suppliers? Because they like firing people for the fun of it? No, because by making something cheaper they can make more money. That edge is very short lived; competitors will do the same, and everyone’s prices will be slashed to restore profits to the previous level. Or prices that otherwise would have gone up to cover increased costs can stay the same. That means that consumers can buy stuff at a lower price.

      What do consumers do with their savings? Do they take those savings dollars and toss them in the furnace? No, they spend the money on other stuff, or save it so they can spend more at a later date. When they spend the money, can suppliers provide that without using any humans? No, when those companies get more business, they probably have to hire people, especially if they are in a service business where productivity per worker has its limits.

      To use Bastiat’s lingo, what is seen is the layoff of the factory worker, what is unseen but is every bit as real as the layoff, is the hiring of someone somewhere to help provide whatever consumers are buying with the money they don’t have to spend anymore on other things because of automation, or cheaper supply overseas. This is why 95+% of farm workers could be automated out of their jobs and still find employment elsewhere.

      True, the older laid off factory worker may not be in a position to take the newly created job as a Pilates instructor in some other town altogether. That problem isn’t easy to solve, and is partially caused by the fact that as society gets wealthier, people get diminishing returns from things, and prefer more services. But service jobs can still pay well, and can be taken by people who never could have qualified for the factory job. They need jobs too.

      It will all work out well in the aggregate, and our standard of living will continue to rise as companies can bring down costs, charge lower prices, and allow consumers to have more of what they want and still have money left over to buy other stuff. We should help those who can’t adapt, but freezing progress to keep everything the same is not the way to go.

      • Wrong Species says:

        If this is the case, then why do people not seem happy about it? Those without jobs report lower levels of happiness, more people are turning to opioids, suicides keep increasing and people yearn for the days when labor unions gave higher wages? That’s certainly not the attitude of people who are content with their financial situation.

        • AnteriorMotive says:

          He says himself, that while technological unemployment isn’t scary in the long run, the process is happening faster than many people can adapt to.

          Technological progress steadily increases the average worker’s living standards, punctuated by the occasional dramatic shock, when they’re the one whose job just got optimized out of existence.

        • Cliff says:

          Is this argument: “The numbers say the economy is doing well, but it can’t be because suicides are increasing”? Couldn’t suicides be increasing for some other reason (e.g. loss of community)?

      • herculesorion says:

        The problem here is that the wages generally fall faster than the prices do…and not all prices fall to the same degree.

        Maybe I can buy a fridge with my welfare check, but I can’t buy my kid a college education or an insulin pump.

    • benwave says:

      This is undoubtedly true, and happening. But at the same time, what about cost disease? The amount of money it costs to house, educate, and provide health care for people has skyrocketed during that time. Of those, at least housing is something that people can’t reasonably live without. This is surely a problem that needs to be solved in order to realise your proposed economy?

      • Cliff says:

        Has housing really increased in cost generally? Taking into account larger size etc.? I thought it was only if you wanted to live in Manhattan, SV etc.

          • Cliff says:

            Correct me if I’m wrong, but isn’t Case-Schiller a survey of 10 or 20 metro areas? And even that shows all the increases over the last 130 years concentrated in just the last 10 years (i.e. prices did not change for 120 years).

  62. AnteriorMotive says:

    Also, I mentioned this the last time it came up, but I think the horse example is deceptive: the bottleneck for horses is human labour. Horse breeders, handlers, etc, found better uses for their labour and transitioned to other sectors of the economy. Put another way, horses were priced out because they need to hire humans to be able to do anything, and humans’ opportunity cost got so high they could no longer afford to outbid it.

    You can tweak the metaphor to make the intended comparison more applicable, but it rapidly starts losing its rhetorical power. I recommend the discussion in the comments of: http://slatestarcodex.com/2017/02/06/notes-from-the-asilomar-conference-on-beneficial-ai/

    “This wasn’t to say that they aren’t an argument for human obsolescence, just that when you exchange the horse analogy for the structurally identical ‘cassette tape’ or ‘slide-rule’ anology, suddenly it’s lost its rhetorical punch. As I see it, this suggests a misleading analogy, hinging on the listener inappropriately anthropomorphizing horses.”

  63. Eponymous says:

    I think it is important to be clear on precisely what the concern is here.

    Clearly in the aggregate, assuming robots don’t kill us, it i to our benefit for them to replace human labor. Humans value leisure, and if robots become sufficiently productive that human labor productivity no longer exceeds the value of our time (which will go up due to our higher income), then we will voluntarily choose not to work. Then follows the post-work nirvana!

    Thus concerns about this process much be linked to distributional considerations. If the robots are owned by a fraction of the population, then it’s quite plausible that the rest of the population will be in for a rough time. Though personally I think it extremely likely that the majority will refuse to accept this outcome, and will enact some sort of basic income through (hopefully) democratic means.

    As to whether this is happening now: it quite clearly is, at least to a limited degree, in some parts of the economy. Autor’s work on the automation of routine tasks shows this. Will the same thing eventually happen to all workers? Assuming general AI is feasible, I think it surely must. How could it not?

  64. deciusbrutus says:

    If robots become available for $10 that cost $1/hr to operate and do everything better than humans, I’ll work for $.50 an hour for a 40-hour week while stealing food, and buy one of those robots; next week I’ll have the robot steal enough food for five people, have all five of them work for cheap and buy a bunch more robots. Once that snowballs into a critical mass, I’ll be able to devote a cadre of robots to solve the FAI problem before the robotic designers design a robot that builds itself for free and pays you to operate it.

    Or in the more modest outlook: Labor force participation isn’t a meaningful metric in a post-labor-scarcity environment.

  65. My support for the basic income is very dependent on this. I’m very strongly convinced that full automation will eventually happen and will lead to mass unemployment, and the need for a basic income guarantee, but this backs up my opinion that it’s not really happening now, and this means that pushing for the BIG now could be counter-productive. If the BIG causes people to not work, and the experiments in various places fail, then this will sap the confidence and support in it for when we truly do need it, when people not working is the reason for it and not an effect of it.

    5. Technology seems poised to disrupt lots of new industries very soon, and could replace humans entirely sometime within the next hundred years. (???)

    I trust the AI researchers, so I’d put a high percentage on the second part even if the lower part deserves a lower percentage. Purely on the logical grounds of accumulation of knowledge about human biology plus time, it has to happen at some point. It’s just that AI researchers think it’s going to happen within that timeframe, or not long after. I think we should trust their diagnosis.

  66. sdenton4 says:

    There’s potentially an error in this caption:

    “Note that, contrary to an extremely pessimistic picture, this would suggest that most people who leave middle-paying jobs go to higher-paying jobs.”

    Given the categorization of jobs by type, there’s presumably a hell of a lot more low paying jobs than high paying jobs. So a smaller percentage increase in low paying jobs could presumably account for a lot more workers than the larger percentage increase in high paying jobs.

  67. Tenacious D says:

    Regarding horses, a book I’m reading now about the environmental history of the middle east (Water On Sand) makes the point that animal power was paramount until oil came into use (coal could fuel ships and trains but animals were needed for smaller-scale transportation). Interestingly, the change was especially dramatic in the middle east since it has a lot of land that is only suitable for grazing (not human crops) so animal labour was relatively cheap into the 20th century.
    So is the decline in working horses technological unemployment or just a change in the predominant source of energy?

  68. Edward Scizorhands says:

    Had we kept 2000-levels of productivity and applied them to 2010-levels of production, we would have required 20.9 million manufacturing workers. Instead, we employed only 12.1 million.

    Productivity got rid of 40% of manufacturing jobs in one decade? Is that right?

    • sharper13 says:

      No, the part you’re not considering is that production (manufacturing output) also increased significantly during the same time period. So if all of that increased output in 2010 required the same number of people as if it was 2000, more would have been hired up to a total of 20.9 million. Of course, the requirement to hire those people would have made the output more marginally expensive and thus gotten rid of most of the increase in output at equilibrium, so it is an odd way to state it.

  69. Squirrel of Doom says:

    The horse example is so utterly dumb.

    I labor market terms, horses weren’t employees. They were tools. They’re analogous to slide rulers, not coal miners.

    It a real well made persuasive trap, which is why it’s spread so wide. We can empathise with these lovable creatures and the hard work they were doing, etc, and never stop to think and analyze. In that sense it is really interesting and worth analysis.

    But as labor market analysis: Pure garbage. Tell your friends!

    • raj says:

      Isn’t labor just a kind of capital? An employee may as well be a tool, albeit one with high upkeep cost and regulations for how it can be used. It’s just that as people, we are concerned with the aggregate demand for people-labor.

      “People aren’t tools” doesn’t address the horse argument at all. Historically people labor has been extremely non-substitutable, like horse labor up until the point of the internal combustion engine. It’s a strong analogy.

      • INH5 says:

        The difference is that, as AnteriorMotive writes above, horses require lots of human labor in the form of trainers and handlers in order to be productive. If the trainers and handlers have better jobs available to them, then the number of usable horses goes down as they move to other industries.

        Imagine that there was a significant population of mutant humans who could understand human language and telepathically communicate with horses but suffered from severe physical and/or mental disabilities that made it hard for them to do anything else productive. Would they be unable to find work in today’s world?

        No. There are plenty of areas of unmet demand for things that they could tell horses to do. Giving rides to children. Carrying heavy equipment for people whose jobs involve traveling in undeveloped terrain that is unsuitable for wheeled vehicles (there’s enough unmet demand in the military for this that they spent millions trying to develop a robotic pack mule, eventually shelving it because it was too loud for combat). Doing the same thing for people on recreational hikes. Helping the police keep order in crowded places. All things that horses do now, of course, but things that could be done a lot more often if trainers and handlers weren’t so hard to come by.

        Look at the prices for this service that provides ponies and handlers for children’s parties, for example. Do you really think that the hypothetical mutant telepathic horse handlers couldn’t significantly reduce those prices, which would in turn expand the business and increase revenue for the business owners, while still earning a decent hourly wage?

      • Controls Freak says:

        Here’s a quick way to test whether something is a tool or a human – ask it, “If we cut your wages 90%, would you decide to do something else?” I’m not sure if you’re from the States, but for just over one-point-five centuries, it’s been illegal to own human beings here. As such, humans are not only capable of analyzing their own opportunity cost and comparative advantage, they’re free to act on that determination all on their own. Horses are neither capable of this nor free to do so. Therefore, they are more like hammers, slide rules, and other tools than they are like humans.

        (Yes, if you work for a company, your boss also makes a calculation of the company’s opportunity cost and comparative advantage, but you get to make yours separately, and he can only ask you whether you’d be willing to mutually agree to a transaction.)

      • Squirrel of Doom says:

        A tool is a thing you own. An employee is another human you have a consensual relationship with.

        Huge and fundamental difference in so many ways that it’s hard to know where to start to explain. One place is that if people stop buying tools, they stop getting manufactured. If employees get laid off, they increase the labor supply, drive down wages and the labor market settles on a new equilibrium.

        I’m trying to understand your “non-substitutable” argument, but it just seems to say that since there used to be a lot of “working” horses, just like there are a lot of working people today, people jobs will disappear just like horse “jobs” did.

        Or, restated: “X used to be a large number, that later went to 0. Y is now a large number, and it will therefore also go to 0”. I’m sorry if this sounds snarky, but I can’t steelman this any better.

    • Edward Scizorhands says:

      THANK YOU.

      If the horse’s owner doesn’t know what to do with it, he shoots it in the head, and the horse doesn’t object until it’s too late.

      The human can adapt. I don’t know how well (that’s what Scott’s whole post is about!) but he won’t sit there blankly until liquidated.

      • pontifex says:

        The human can adapt. I don’t know how well (that’s what Scott’s whole post is about!) but he won’t sit there blankly until liquidated.Report

        You never worked in retail, did you?

        • A Definite Beta Guy says:

          Yeah, there are a lot of humans who are REALLY dumb and REALLY lazy.

          These humans are theoretically more capable than the horse, and probably are for most things, but I could imagine some rather rare scenarios where the horse outsmarts them through sheer persistence. Maybe not in chess, but in “hey, there’s a bucket of water, I need that bucket of water, how do I get that bucket of water.”

  70. habu71 says:

    I have learned to love the words “Much more than you wanted to know”.

  71. pinyaka says:

    I don’t think it would take self delusion for your coal-miner turned househusband to claim they’re not looking for work. The guy looks for work, sees that what’s available is more appropriate for his wife (for whatever reason) and they move to the situation you describe where she works and he doesn’t. He knows that there’s nothing available for him so he doesn’t look. Possibly if a new mining company opened up he’d go apply for it, but it’s not delusional for him to accept the current state of affairs even though he’d totally jump at a chance for a better one.

  72. Garrett says:

    Two things to add:

    1) Part of the disability growth is likely (in part) due to incentives between governments. IIRC, the majority of the cost of unemployment benefits lands on the States, but the majority of the cost of disability benefits lands on Federal Government. NPR Planet Money Episode about this very thing.

    2) Part of the issue is that for some jobs we have story for how automation will work, and for others we don’t. If you go to start a road paving business, you know you are going to be using heavy machinery like backhoes, dump trucks, etc. You could, in theory, have people use shovels and picks, but that would be so inefficient that it isn’t considered.

    On the other end, if you go to start a Florist business, there’s no way of automating most of the specific work. What would it even mean to go to have an “automated florist”. Sure – some of the labor like cash-handling and delivery can be automated, but the core elements like floral arranging and consulting aren’t.

    Maybe in the future there will be a way to automate away the jobs. But that won’t happen until we have general agreement on how it should work.

  73. pinyaka says:

    I don’t necessarily think that most people think that technological unemployment will hit the least skilled first. It should hit the jobs where it’s most profitable first. How cheap would a robotic server in a restaurant have to be to actually get a return on investment? Replacing a unionized assembly line worker makes you money a lot faster even if it’s more expensive. As we get better at replacing middle cost jobs, the cost of automation should go down and that should allow us to replace the unskilled.

  74. glorkvorn says:

    Great article! I remember reading a lot of this sort of stuff during the Great Recession 2009-2011ish (when I was also unemployed). It seems like there’s not as much interest in the subject anymore now that unemployment is back down, but I’m glad people are still researching it.

    “The argument against: we’ve had increasing technology for centuries now, people have been predicting that technology will put them out of work since the Luddites, and it’s never come true.”

    Are we sure that’s actually true? My vague impression of the past (wild generalization about all of human history) is that there was never-ending toil, just ridiculous amounts of work to be done. So much so that most farmers needed like 5 kids just to keep things running, and even that wasn’t enough to do everything they would have wanted. So they just had to accept that not everything would get done, and lived with bad sanitation and less productive fields. “Unemployment” was basically non-existant, since even the very worst man- a wandering homeless drunk, say- would still be useful for *something*. The technological unemployment has come in the form of reducing labor to what humans can feasibly do, rather than this never-ending pile of work that’s far too much for anyone.

    “All of this is a mess. But the impression I get from this mess is that there is little sign of technological unemployment happening today. I get this from a few sources.”

    Are we sure that we can actually trust all of this data? What if this is a “people are lying on the internet” thing? Like your hypothetical laid-off West Virginian factory worker- he might not want to admit to some nosy census-taker that he’s been laid off and is now given up or a homemaker or whatever. He might also be illegal things to make money- disability fraud, or just straight up working in the black market. It’s notoriously hard to get good social science data, so when different sources of it seem to contradict each other like this I’m suspicious that one or more is simply wrong.

    • ilikekittycat says:

      The idea of the past as never-ending toil was a Victorian era invention to justify their 60+ hour work weeks. We work more in industrial capitalist times. A thirteenth century agricultural serf had to work very grueling 12 hour days for ~150 days a year. There was lots of time for pilgrimage, jubilee, local festivals, leisure, etc. the rest of the year. The ancien règime in France, to give one example, guaranteed fifty-two Sundays, ninety rest days, and thirty-eight holidays.

  75. M_Aberson says:

    I think a few worthy notes are missing from the front-end of this analysis:

    1. Stripping out male prime-age LPR as the clear trend of evaluation is problematic. Women entering the labor force is such a massive shift in how the labor market operated. The demand-side effects of women entering the labor market is far more meaningful in understanding the pre-recession drop in male PALFPR than anything else. Total PALFPR seems like a much better metric to use and it has been reasonable close to what you would expect from the “creative destruction” story. It peaked at 84.6% in the late 90s at 84.6%, bottomed out in 2013 at 80.6% and has been steadily climbing since then and now stands at 81.8%.

    Prime Age LFPR

    Wrapped up in women LFP process is a pretty good story on why low education male labor participation would decline relative to highly educated men. They were more likely to have a spouse with better career prospects that could support than they would have been in the past.

    2. There was a distinct and significant change to disability eligibility in the Social Security Disability Reform Act of 1984 where pain was introduced as a qualifying criterion.

    established by medically acceptable clinical or laboratory diagnostic techniques, which show the existence of a medical impairment that results from anatomical, physiological, or psychological abnormalities which could reasonably be expected to produce the pain or other symptoms alleged

    NCBI Paper on the subject

    This was a significant shift in the accessibility of SSDI that could potentially explain a great deal of the increase in disability as a reason to leave the labor force without considering a supply-side shift.

  76. Snailprincess says:

    I feel like we more or less know what’s causing ‘wage stagnation’. It seems to be largely a function of measuring ‘wages’ and not total compensation. Here’s a link with some data: https://www.heritage.org/jobs-and-labor/report/productivity-and-compensation-growing-together

    If you measure total compensation (wages + benefits) then compensation hasn’t stagnated at all, it’s continued to rise at about 1% year (just like it has for a long time). It’s just that we’re pumping those increasing consistently into increased health care costs. It makes sense that this would be most apparent in the middle income range. Low income people likely don’t receive medical benefits, or at least not comprehensive ones, high income people make enough that increased medical costs don’t eat all of their wage increases. It’s the people in the middle who end up seeing virtually all of their increased compensation get eaten by increased health care costs.

    The numbers on this seem pretty clear. There is no ‘total compensation stagnation’ just ‘wage stagnation’.

    • mobile says:

      Ditto for ‘individual compensation stagnation’ and ‘household compensation stagnation’.

    • Wrong Species says:

      What about cost disease? If health care costs keep increasing dramatically without a corresponding increase in quality, that would mean people’s total real compensation has decreased.

      • Snailprincess says:

        I would label that as true but unhelpful. If my income is growing, but I’m enjoying it less because I put an ever larger percentage of it in the blender, then the solution is not ‘make more money’ but ‘stop putting it in the blender’.

        • John Schilling says:

          Not putting money into the blender is such an obvious solution that we should be skeptical of claims that the thing people are putting money into is really a “blender”. Until we understand what is causing cost disease, and per our last discussion here we don’t, we shouldn’t assume that it is a thing we can stop without drastic consequences.

          • Cliff says:

            There seems to be strong evidence that healthcare does not reduce mortality and also quite strong evidence that nutrition and exercise strongly influence mortality.

  77. mobile says:

    One argument that I don’t see and am quite surprised wasn’t made by any economists cited is the relationship between wealth and labor force participation. We are much wealthier than we were 50 years ago, and we also work fewer hours than we did 50 years ago, not to mention the era before “the weekend”. We live longer and retire earlier. Child labor, aka the pre-prime age male labor force participation rate has plummeted to virtually nothing, but that’s because our general prosperity has made such labor unnecessary and we think it is wiser for children to invest their time in compulsory schooling rather than productive work, and not because any new-fangled machinery does that work that 12 year olds used to do, only cheaper. Looking forward, it’s not hard to find economists who predict that many people in a generation or two will have 15-hour work weeks.

    Surely this affects labor participation rates at the margin, especially if the social safety net is stronger and/or today’s low-cost leisure activities (e.g., video games) are more compelling to prime age men. It would be interesting to see what these charts look like for western Europe, with the same general adoption of workforce technology and prosperity trends, but stronger safety net. I would predict that the trends you cite that appear to show technological unemployment would be even stronger in Europe.

    (Apologies if any other commenters made this point; I just quickly scanned the comments for some relevant keywords)

    • Lasagna says:

      We are much wealthier than we were 50 years ago, and we also work fewer hours than we did 50 years ago.

      I’m not being hostile, but my gut is that both of these assertions are untrue. What makes you think they’re accurate?

      What’s giving me pause is:

      1. 50 years ago, a person working a blue collar job could afford to own his own home and support a family on just the one income. Now, they’d be on the street. In what way are we wealthier?

      2. Who is working fewer hours? 9-5 was the standard when my grandfather came home from the war. I’ve never worked a 9-5 job in my life (unless you count when I worked at McDonald’s, where the shift was about that long), and I’m 45. I’ve worked 60-80 weeks until about five years ago, when I was blessed with a 50 hour a week job. I never expect to work less than that. And my work hours seem pretty typical.

      • Brad says:

        1. 50 years ago, a person working a blue collar job could afford to own his own home and support a family on just the one income. Now, they’d be on the street. In what way are we wealthier?

        I can’t quite wrap my head around the fact that anyone would think that was a good way to measure the wealth of a society. Maybe to raise questions about how wealth gains (if any) are distributed, but not as a direct measurement of wealth.

        Also, it isn’t true. At least not without some qualifications. There are plenty of blue collar jobs paying enough to own a home, in much of the country, and to support a family. Particularly if that family has material needs and wants equivalent to those of a family 50 years ago. And if you argue that they don’t and shouldn’t, well there’s your wealth increase.

        • Lasagna says:

          I can’t quite wrap my head around the fact that anyone would think that was a good way to measure the wealth of a society.

          Sorry – can’t help you, then. Seems fine to me.

          Also, it isn’t true. At least not without some qualifications. There are plenty of blue collar jobs paying enough to own a home, in much of the country, and to support a family.

          No kidding! Here’s what I want you to do: provide examples of jobs that receive an hourly wage which pay enough to own a home and support a spouse and children. The “material needs and wants” are equivalent to a family 50 years ago – beyond the house, let’s throw in a car and a TV.

          • Brad says:

            http://flcdatacenter.com/OesQuickResults.aspx?code=37-3013&area=15380&year=18&source=1

            In the Buffalo, NY area tree trimmers have a mean wage of $46,966 a year. For a family of four with that income, taxes are less than $5000. For a net income of $41,966.

            Here’s a 3 bedroom, 1 bath house in Buffalo for $108,500. https://www.zillow.com/homedetails/78-Altruria-St-Buffalo-NY-14220/30216554_zpid/

            With a 20% down payment that implies mortgage payments of around $450/month. Call it $800/month all-in for housing. That’s $9600 a year. Leaving $32,366 a year or more than $2500 a month.

            Is it your contention that $2500 a month after housing isn’t enough for a 1970 standard of living for a family of four?

          • JayT says:

            You could be a garbage man in Oakland ($61,000 a year, which, from my experience, actually seems low to me) and buy a home in Oakland for $300K.

            Cars are much more expensive today than they were 50 years ago, but they also have many more features, many of which are required by the government. They are also much more reliable, so I would say that a used car is more comparable to a car from 50 years ago than a new one. Even then a car that’s a few years old will be much better than a car from 1968. Let’s say that our garbage man spends $15K on his used car, which works out to ~$3,000 a year if he finances it.

            He can buy a TV that is far better than anything he could buy in 1968 for $100.

            After taxes, mortgage, and his car that leaves him with about $30,000 a year for all his other costs. If he really did live like someone from 1968, he could probably enjoy a lifestyle very similar to a middle class person from that time.

          • Cliff says:

            Blue collar jobs are not typically very low income. Retail jobs are. Welders, mechanics, loggers, fishermen, tradesmen, etc. make decent to very good money.

      • Controls Freak says:

        In what way are we wealthier?

        Real gross domestic product per capita is a good start.

        Who is working fewer hours?

        Uh, people. You may not be the average, but we all make different choices/tradeoffs.

    • Swami says:

      Yeah, I agree, Mobile.

      Incomes are up considerably, especially when adjusted for benefits, transfers, taxes, and demographic household transitions. Leisure and retirement is up, continuing a hundred year plus improving trend. I think the non participation rate better explained by rising affluence and transfers than by stagnation, though is wages rose even faster it might counteract things a bit (changing the costs and benefits of not working).

  78. dalemannes says:

    “If you’re very conservative, you might say – aha, I knew that people were just unemployed because they’re lazy! But if you’re more progressive, you might ask – exactly how miserable do you have to be before you stop working?”

    Wow. What a ridiculous villain caricature of conservatives and a heroic portrait of progressives. Normal conservatives (as well as progressives) genuinely sympathize with manual labor workers with health issues and the desire for dignified work.

  79. romeostevens says:

    And after all you’ve seen in all this time, why does your heart still crave a ‘educational, social, or political intervention’?

  80. DocKaon says:

    The Bessen (2015) ATM study is a great example of the conclusion not at all matching what the actual data says. Go look at the plots in that paper. What I see is that 10 years after ATMs are introduced employment of tellers plateaus and stays there with very little increase until the present. Virtually all the job growth in the sector occurs within the first 10 years after ATMs are introduced. Eyeballing it, from 1970-1980 tellers increase by ~5% a year and 1980-2010 by 0.6% a year. In other words, it takes approximately 10 years for the ATMs to become widespread and impact tellers jobs.

    Maybe it doesn’t fit some cartoonish view of how technological unemployment works, but if an ATM equivalent was introduced for my industry I would certainly be scared.

  81. Mwncsc says:

    Suppose there is near-universal technological unemployment because of perfect androids and that the rich control these and that there is no UBI. Wouldn’t that just create two economies?

    On one level, you’d have the wealthy on their estates and their robot armies churning out goods for other wealthy people. But there would still be a massive untapped demand from the other 99% of people. Wouldn’t some of those enterprising 99% start an economy of their own based on labor for other people without robots (who still would need goods and services)? Wouldn’t there be a renewed trade in whatever goods people could make by hand or with simple tools from what resources were available? This sounds like a pretty bleak kind of techno-feudalism, but if things made by the robots (and indeed the robots themselves) are so cheap and plentiful, then one wouldn’t have to generate much value in order to upgrade to a robot operation.

    Or am I missing the part where the rich liquidate everyone else en masse?

    • angularangel says:

      Hmm. I wrote a response, but it’s not showing up, even after I posted it again. It was pretty long, maybe it ran afoul of an auto-moderation feature? :/

      • angularangel says:

        Theeeere we go. I just had to break it in half first before it would go through. Little long, I know, but this is slate star codex. 😛

    • angularangel says:

      I personally would reference Meditations on Moloch for such a scenario. What happens in a competition when some participants get an insurmountable advantage and others don’t? The ones who didn’t lose. In this case, losing probably means falling as far economically as you’re allowed to.

      That might just mean sitting back on your couch and collecting an U.B.I. while the rich travel to space or whatever – or it might mean finding that you can no longer afford to survive. If, for anything you can make, there’s a machine produced equivalent thats better and 1/100 the price… what are you to do? About the best outcome I can see, failing societal or government intervention, is that you own some sort of property with which to sustain yourself. Land, stocks, whatever.

      But thats not something that will apply to everyone, and it won’t be a long term solution. Unless you’re really lucky (I.E. one of the rich), your property won’t be enough to sustain you forever, or your kids, or your kids kids… Eventually the money runs out. So then you’re homeless and have no food. Squatting and begging is the next recourse, but that basically depends on the wealthy. If they decide they don’t want you doing that, they’re gonna send their robot police to tell you to quit it, you’re gonna be out of luck.

      So what then? Well, your best bet is probably gonna be being sent to prison for squatting or whatever. It’s either that or a horribly overcrowded shanty-town on whatever public land remains, and honestly, you’re probably better off with the prison. And thats assuming you don’t just get killed by the robot police or other unfortunates first. So, yes, essentially. I think the rich will liquidate everyone en mass.

      Honestly, I think this is already happening, only very slowly and practically unnoticeably. It’s mostly the slow erosion of property, accumulation of debt, slow grinding of people into homelessness, and ever growing wealth and power of the rich.

      Now, there are some caveats – some people will manage to tag along with the rich – their relatives, their friends, those offering their novelty as human beings. But I don’t think that will be a very large portion of the population. 10%, tops.

    • angularangel says:

      Now, yes, that is the worst case scenario. There IS the potential for government or societal intervention. We could implement an U.B.I., or a jobs program, or systematic neofeudal LARPing, or whatever. But on this, too, I’m skeptical. I think there’s good odds we could institute something, but I don’t think it would last. Unless it was hefty enough to grind the rich back down to scale with the rest of us, and prevent any other accumulation of power and competitive advantage, it would be a purely palliative measure.

      In time, the rich would grow tired of it, and begin to wonder “Why are we paying these people anyway? They’re basically just parasites. They do nothing useful.” (They’d even technically be right.) So they start looking to cut the benefits, cut the numbers of people on the dole, cut the money they personally have to put in to it, whatever. Take all that money spent on education – is that really necessary? They’re not going to work anyway. Lets cut that. What about all that money spent on healthcare in the last 30 years of life? If we just have them die early, we can save all of that. What about population growth? If we just cut that, we might be able to stop paying out entirely in a century or two!

      By methods like this, the population can be reduced, until the rich no longer need spend anything on the poor at all. So, now that Moloch has claimed his feast, surely now we’re done, right? Surely now those who remain can live their lives in peace and happiness?

      No, not necessarily. Moloch is a gluttonous god, he is not so easily satisfied. Now, if not earlier, he turns his eye to the rich and looks to play them against each other. Will they compete economically? Politically? Will they turn their machine armies against each other? Will they race for more resources, to plunder and pillage first, and thereby amass more wealth? Remember, it only takes a couple who wish to compete, to pull the entire collection back into the struggle. And yes, they can always cooperate, and threaten violence against any who try such things… But this is not perfect. We have plenty of evidence showing how such measures can be defeated. There will, most likely, still be competition, and as such there will still be winners and losers, and accumulation of power and competitive advantage. And as such, Moloch will continue to receive his due.

      But okay, that’s enough pessimism. Things might not turn out that bad. Perhaps the rich will be satisfied with their previously-unthinkable wealth, and won’t begrudge the rest of us a dole, or perhaps even we’ll find that we all finally come together in an embrace of our mutual humanity and everyone sings kumbayah happily ever after. Perhaps we’ll become aware enough of this kind of problem, and solve it. Perhaps we’ll have our culture/government/whatever lock the rich into some “perverse failure to optimize” that works out in our favor. Perhaps the rush of new resources will be so great, it’ll buffer us against such things for several thousand years, or even forever. Perhaps we’ll get lucky and find that the things we like and care about are, in fact, optimal, and are selected for, not against. Perhaps we’ll advance into glorious posthumans and all of my theorizing will be irrelevant. Who knows. But, as things are, I’m pretty skeptical.

  82. Dacyn says:

    – “There may be some point at which we too stop being worth more than it costs to replace us.”
    I think you mean “at which we too stop being worth more than we cost”? Or “at which we cost more than it costs to replace us”? It doesn’t make sense to compare the two things that you’re currently comparing.

  83. JulieK says:

    Most people expect that technological unemployment will hit the least skilled first

    The jobs that have been eliminated in the past by technological unemployment were often skilled or semi-skilled ones, like scribe or weaver.

  84. VolumeWarrior says:

    Surprised to see no mention of videogames. The career prospects of young men are humiliating or tediously circuitous to puruse. So they just stay at home and play xbox.

    http://marginalrevolution.com/marginalrevolution/2016/09/labor-force-participation-video-games.html

    http://marginalrevolution.com/marginalrevolution/2016/07/what-are-young-men-doing.html

    • pontifex says:

      The solution is to make video games humiliating and tediously circuitous to pursue. World of Warcraft is ahead of the game here — “fetch 99 rat pelts to advance to the next level”, etc.

      • moonfirestorm says:

        WoW’s moved pretty far beyond “fetch 99 rat pelts”. In your normal leveling experience, it’s extremely rare that you’ll be on the same quest for more than 10 minutes of time spent on it, and they’ve balanced drop rates so that usually if you have 2-3 quests in an area you can get them all done at the same time. Grinding is a thing of the past, although you do sometimes run into it if you want the old rewards.

        The Artifact Power system has replaced that to some extent, but only at endgame (when you’re looking for something to do anyway) and it naturally inflates away progress, so there’s only as much grinding as you want to do. Imagine if in the first week it took 99 rat pelts to get something, but 2 weeks later it was only 30 rat pelts, and eventually it was just 1. It can get tedious, but only if you push it way beyond where a regular player would (for extremely limited payoff), in which case you sort of bring it on yourself.

        WoW’s new thing is the Skinner box (or rather, bringing it more to the forefront of the game): “do this thing lots of times, and sometimes you’ll get a legendary/Titanforged piece/etc”

  85. JulieK says:

    There was a type of employee at the beginning of the Industrial Revolution whose job and livelihood largely vanished in the early twentieth century. This was the horse.

    There’s a type of employee that’s vanished between the early 20th century and now- the pre-teen.

    At the beginning of Anne of Green Gables, a middle-aged couple is planning to adopt a boy around 10 or 11 years old to help them on their farm. When the orphanage sends them a girl by mistake, a neighbor offers to take her off their hands, since she can use help with the housework.

    How many households nowadays think an 11-year-old would be an immediate economic asset?

  86. JulieK says:

    In 1970, educated and uneducated men were about equally likely to be PAMLFNPers. The rate for educated men didn’t change. The rate for uneducated men shot up.

    Fifty or 100 years ago, knowing that someone had only a high school education or less didn’t necessarily tell you much about his intelligence, conscientiousness and so on. More likely it meant that he came from a poor family, and left school as soon as possible to help his parents put bread on the table.
    Since then, our society has gotten more and more effective (or ruthless) at sorting people into different tracks. Thus, comparing “people with X level education in 2018” to the equivalent group in the past is comparing apples and oranges.

    • proyas says:

      Olaf Stapledon predicted this in his 1942 book “Darkness and the Light.”

      “Within a few generations this policy of fostering intelligence and integrity began to have surprising results. Society began to be stratified in ranks of ability. People tended to confine their mating within their own rank of capacity. Consequently the first signs of a new caste system appeared.”

  87. proyas says:

    The increase in U.S. obesity rates since 1980 could explain the rising rates of disability-related nonemployment and nonemployment due to disinterest, especially when you remember there’s a strong link between obesity and depression.
    https://stateofobesity.org/obesity-rates-trends-overview/
    https://www.cdc.gov/nchs/products/databriefs/db167.htm

    Also, it might be the case that the entry of women into the workforce starting 50 years ago permanently depressed wages and drove down the labor force participation rate. Is there an economic theory stating that the number of jobs must keep growing and their wages must also grow or stay the same, regardless of how many new workers are added to the system?

    • albatross11 says:

      It seems like one effect of entry of women into the workplace was that a lot of work that was formerly not included in economic statistics got included.

      Suppose I’m staying home with my kids and taking care of them. Then, I get a job and put my kids in daycare. The official economic statistics will correctly record the extra economic activity of my job, but they will also record the extra economic activity of someone getting paid for watching my kids, even though that’s a job that was previously being done but not recorded. (The country is no better off when I’m paying someone else to watch my kids than when I’m watching them myself!)

      So it seems like the economic statistics should be overstating how much the economy grew during the time when women entered the workforce in a big way. They’ll show both the real increase (the woman doing some job that pays a salary) and an on-paper-only increase (someone being paid to do the work she was previously doing without any formal paycheck).

      • proyas says:

        That’s a good point, though the new jobs would be low-pay occupations like babysitter and food service (working mom gets take-out from McDonald’s because she doesn’t have time to cook). That says little to nothing about how the mass entry of women into the workforce might have affected the job market in middle-income positions where men used to dominate.

        Picking a random example, if the number of people with J.D.’s in America doubles because women are no longer barred from entry, is there any reason to assume that the economy will automatically expand to absorb them all, with the number of attorney jobs doubling and salaries staying the same? “Is there an economic theory” that says such?

        • albatross11 says:

          If the supply of JDs goes up (the supply curve shifts right), then we should expect to see, short-term, more JDs employed at somewhat lower wages. This is what pretty-much always happens when the supply goes up. Over time, I think we’ll see two other things happen:

          a. Lots of people with JDs will take more standard legal jobs that are currently taken by people with normal law degrees. If JDs become common enough, it may even become normal to require a JD for some job that previously required only an ordinary law degree, just as has happened with requiring a college degree for jobs that previously required only a high school diploma.

          b. If there are some tasks which were previously not done because there weren’t enough people available with JD-level knowledge of the law (so it was too expensive to hire the huge staff of legal scholars you’d need to carry out some new line of business), then those businesses will become viable as the cost of JDs goes down and their availability comes up. (But I don’t know enough about law or the legal profession to know whether this is a realistic scenario in practice.)

          But wouldn’t the total number of JDs have actually remained about the same (or perhaps the total number of high-prestige law degrees–say from top ten law schools)? I’d expect the change to be that more of those JDs (MDs, PhDs, DVMs, etc.) are women.

    • proyas says:

      ‘Although an obesity diagnosis alone is not enough to qualify for disability benefits, there are circumstances under which an obese person may meet Social Security disability medical eligibility requirements. These include cases where a person’s BMI is so high that they are unable to move, walk, or complete everyday tasks like preparing food, cleaning their home, or dressing or bathing.

      Obese people may also qualify for disability benefits by meeting or matching a disability listing for a related medical condition, like heart disease, joint disorders, diabetes, or stroke.

      If you are unable to work due to your obesity or the health complications your body weight causes, then SSD benefits can help pay everyday living expenses and medical bills. Social Security disability qualification additionally makes coverage through other state and federal benefit programs available to you, potentially including Supplemental Nutritional Assistance (SNAP), Medicaid, and Medicare.’
      https://www.disability-benefits-help.org/disabling-conditions/obesity-and-social-security-disability

  88. JulieK says:

    You compare the stories of education-as-necessity and education-as-credentialism.
    How about a third story – education as a luxury? With our society becoming more and more prosperous, we can afford to let our children spend more time in school.

    • albatross11 says:

      Yes. Also, education as a kind of jobs program–we hire more teachers, and we keep people off the job market longer, lowering unemployment in both cases. I don’t know how important that effect is, but it’s probably at least somewhat relevant.

  89. A Definite Beta Guy says:

    I really don’t have much to add. IANAE, but I did study it and do dabble in it, and associate with a lot of similar people. The article pretty much summarizes the consensus view: technological unemployment is a dramatically over-hyped phenomenon, and for the most part you are talking about different returns to different sets of workers depending on skill-sets, with a major hollowing out of routine and semi-skilled tasks.

    To give an example, you can get an overview of some of my middle-skill accounting office work here: http://slatestarcodex.com/2018/02/14/open-thread-95-25/#comment-601463
    You can probably see a lot of opportunities for automation, that would eliminate a lot of decent middle-paying jobs. You can imagine how many routine jobs would have existed even 20 years ago. Like, I can easily review every single client in my entire portfolio and see their outstanding balance, within minutes. Can you imagine doing that prior to the existence of computers?

    So, yes, technological change exists and can eliminate certain positions, including positions that are decently well-paid. The commenter above talking about changes in Big Law describes this perfectly: extremely expensive Ivy League grads who reviewed contracts have been replaced by computers who can do the same damn thing.

    But the idea of widespread increases in unemployment as a result of technological change? No. Doesn’t make sense. There’s always something humans will have a comparative advantage at, and therefore humans will always be employed. There will be some sort of “natural” rate of unemployment and some sort of “natural” rate of labor force participation rate, but these are influenced by government policy and culture.

    The way technological change might work is in conjunction with other policy changes, and Scott already touches on this with SSDI. A worker who had a high-wage occupation eliminated might decide to claim disability rather than work a crappy low-wage job. Or stay with parents. Or pick up a Sugar Momma. But it’s not a STRAIGHT technological unemployment story: he still could be doing something, even if it is just slinging burgers at McDonald’s.

    We might eventually get widespread technological unemployment because an android is strictly dominant, but if that happens, you are talking MAJOR changes. The only change as big in human history was the Agricultural Revolution, so this is equivalent to asking Paleolithic hunter-gatherers to predict the machinations of the Cold War: utterly and embarrassingly futile.

  90. IronEconomist says:

    Firstly a minor point: At the beginning of section 2 the discontinuity that you are seeing is just the generalised fall in unemployment due to the Great Financial Crisis, it is not present in the second graph as that is as a fraction of total employment which also fell commensurately.

    I think that the most favoured theory of wage stagnation among the middle classes among economists is known colloquially as the Great Factor-Price Equalisation. In essence, as globalisation proceeded different sectors became integrated, after integration there was pressure on wages in india, china and the developed world to converge in those sectors. This began in manufacturing and then china and india upskilled into administrative tasks. Those countries are still behind in the top quintile of jobs: top tier software engineering, upper mananagement, etc so those jobs saw no wage stagnation, and actually saw a premium because there was more demand for their services since there were more middle-skilled people who could benefit from their expertise, and because there was more demand for goods generally from the new middle class which makes the upper management of a global company more important. See for example for a very brief introduction

    Expect this not to persist, for the indians and the chinese middle classes to achieve wage equalisation within 15-50 years and the segmentation of the US labour force to end after.

    In short, I think that most of the stylised facts above have quite pedestrian explanations that have nothing to do with robots.

  91. Floccina says:

    If we had as many horses (and could not kill and eat them) as there are people we would find a use for them.
    Robots make things cheap so a person will not need to work so much.
    IMHO It’s the rate of change that makes for the problem, people can compete with robots if need be but if jobs are eliminated faster than new jobs are created unemployment will be the result but that will usually ebb and flow.

  92. Tarpitz says:

    I wonder if part of the poor productivity gains mystery can basically be explained by increased dossing. If improved software and online materials for research enable an office worker to do the work that 20 years ago would have taken them 40 hours in 20, but they’re still required to spend 40 hours in the office and don’t have much incentive to do more or better work, the productivity gains are liable to vanish into Facebook (and indeed SSC). Anecdotally, this kind of statistically invisible underemployment has been pretty common at a number of offices I’ve worked in.

    • anonymousskimmer says:

      I’m skeptical that the office workers of earlier eras didn’t waste just as much time chatting at the water cooler or listening to the radio.

  93. alef says:

    So in 1950 we introduced (or rather, used far more widely than just its former niches) a new technology that was an essentially perfect substitute for manpower. To pretty much exactly the extent it became increasingly deployed, the ‘workforce’ declined (an almost perfect offset for 25 years, a bit looser thereafter). In the past, new technologies have (after some painful adjustment time) lead to new jobs, new opportunities, and a recovery or gain in ’employment’; but in this case we’ve waited a couple of generations and still seen absolutely nothing in the way of recovering from this offset.
    Is this experience not relevant, as robots and AI will very shortly do this again? For a while, they won’t be remotely as capable or flexible substitutes for manpower as in the former case; but they will also be cheaper.

  94. encharitimone says:

    As with so many complicated problems, I think some of the analytical difficulty here comes from horrible, horrible word usage conventions. In this case, “automate”.

    I don’t know what range of definitions people would give, put on the spot, but the definition that fits the usage I tend to hear is something like “remove the component of a task that requires the most human effort”. Depending on the particulars of the task, this can mean wildly different things, and can even result in things being “automated” several times.

    Suppose one day your neighbor excitedly tells you ze just heard about a new automated way to mow your lawn. What do you picture?

    Well, as a member of the educated elite of the 19th century (why else did you think you were talking to your neighbors, not people on the internet?), you probably pictured a push mower (https://en.wikipedia.org/wiki/Lawn_mower#/media/File:ReelMower.png). Swinging a scythe repeatedly at just the right height to get an even cut was a ridiculous amount of work to (pay someone to) do. A push mower automates the hard work of mowing: cutting by using the effort to push it forward to propel the blade.

    On the other hand, if you were a wealthy citizen of the 1930’s, you might know about the first motorized (what we would now call a riding, or self-propelled, at least in the US) mower. It automates the hard work of mowing: pushing the machine around.

    Of course, in reality you are a tech-savy modern, and your neighbor is your coworker in the next cubicle, you likely know that there are now Roomba’s (and competitors) for mowing lawns, that you set out and start, and your lawn gets mowed. This removes the hard work of mowing: being outside in the heat to actually guide the mower so it mows your grass and not your flower bed.

    If companies ever find a way to convince us that smart homes are anything but a huge inconvenience, mowing the lawn will once again be automated: the house will detect that the lawn has grown beyond the programmed optimum, and active the Roomba to go fix the problem, and you’ll come home from work to a fresh-cut lawn. This removes the hard work of mowing: setting out and starting the Roomba.

    It might be more useful to describe the first several steps as “mechanizing” and the latter steps as “automating”, but that’s not how we use the word automate. It’s similar to the “AI” vs “AGI” problem.

    Note also the impact at each stage on a lawn care business: at each stage, the productivity went up, but so did the capital costs. A group of unskilled workers could be competitive with just a scythe and a rake, then they needed push-mechanical hardware (but could do more area), then they needed mowers (but they could do a lot of area), and in a few years they’ll need a fleet of robots. In reality, of course, people who pay lawn workers are usually paying for more advanced landscaping than robots (currently) do, but the idea remains.

    Apply this type of concept to other areas, and I think it’s perfectly reasonable to say, in common parlance, “we have been automating things since the 60’s” and “automation recently started destroying jobs”. Eventually, the “the component of a task that requires the most human effort” is “the last component of a task that requires human effort”.

  95. spinystellate says:

    Suppose that technology over some time horizon pushed 65% of people into jobs that pay twice as much, and 35% of people into jobs that pay half as much. We would say that most people were better off, and that GDP was substantially increased. Most bulk statistics would show us to be much better off. But we’d still have 35% of people who were super pissed and motivate to vote accordingly.

  96. MB says:

    The main difference is that the horses didn’t get a vote. Suppose that policemen’s, soldiers’, secret service agents’, etc. pensions become worthless due to economic upheaval/unkept promises. Suppose their children don’t get to join the new elite or even a first step up the ladder of social advancement, because “there are no more middle-class jobs”. What then? They probably won’t go gently into the night. And if they have a say over these new developments, things will turn out quite differently.
    If the Industrial Revolution created a lot of manufacturing jobs and this new AI Revolution is taking them away, then my prediction is that there will be a return to the previous situation. Many jobs have disappeared in advanced countries because they don’t make any sense economically, but persist in other parts of the world, where labor is sufficiently cheap. Compare the kinds of jobs that remain in Japan with those that still exist in third-world countries, but have been automated in Japan.
    Right now, only truly rich people can afford personal servants and full-time bodyguards. But what about in the future?
    With human labor becoming even cheaper, rich or powerful people will again hire large retinues, with courtiers (poor relatives), bodyguards, and personal servants. Even the servants will have servants, too. Why not hire a couple of servants for each room of one’s palace? Why not keep a craftsman or technician in-house for each conceivable craft or technique one might need? Why not keep a chaplain or two? An in-house tutor for your children? An economist to manage your estate? Or an in-house psychologist? More realistically, why not keep a harem?
    Also, the state may maintain a large bureaucracy, large security services, and a large standing army, as a sort of bribe for whatever is left of the middle class (and because the new servant class will need a lot of surveillance and repression). And all these people will compete in hiring the largest possible number of servants they can afford — the true mark of high social status and success.
    This future sounds a lot like Saudi Arabia.

  97. phoenixy says:

    Have you listened to the fantastic This American Life piece, Trends With Benefits? They do a lot of reporting that includes analysis of trends, but also interviews with people on disability, disability administrators, etc. and come to some of the same conclusions that you do.

    For example, here’s a quote from the piece:

    And it wasn’t just Scott [who went on disability after the mill closed down]. I talked to a bunch of mill guys who took this path. One who shattered the bone in his ankle and his leg, one with diabetes, another with a heart attack. When the mill shut down, they all went on disability. And they all, like Scott, would prefer to be at the mill. Scott says if circumstances had been different, if the entire economy had not changed underneath his feet and the mill was still open, he’d be there.

  98. actinide meta says:

    A problem that I think is underappreciated is that it’s widely considered immoral to trade with people in the lower part of the income distribution.

    If I start a software company that employs, say, 500 software engineers and makes lots of money, I’m a hero. My employees like their jobs, the press writes articles about how extravagant their benefits are, my investors like their money, the government likes their taxes (but otherwise mostly leaves me alone), and most people assume I came by my big pile of money honestly.

    If I start a company that employs, say, 100,000 unskilled laborers and makes lots of money, I’m a villain. My employees probably resent me, the press writes articles about how terrible their benefits are, I have to worry about all kinds of legal risks and regulations, and most people assume that I got my money by taking advantage of my employees.

    I feel like most of the people smart enough to build successful companies are smart enough to do the math on these options and decide that, all else equal, they would rather build the first kind of company. And unsurprisingly we see a lot of new companies in the first category being created. People want “job creation” in the abstract, but when they see it concretely, they want to spit on the people doing it. And entrepreneurial talent, energy, and credibility are limiting resources in economic growth.

    • Worley says:

      I think your comparison is not completely correct. If you hired 100,000 unskilled laborers, but gave them an income that was substantially higher than they had before (and that is higher than their parents had and higher than the people they associate with socially), your employees would love you.

      Where it gets ugly is when the market pay rate for a class of workers is declining strongly. Then if you hire a bunch of them for 10% more than their current market value, it can still be 50% less than they think they’re “worth”, because that’s what they made 10 years ago.

  99. noahyetter says:

    “It’s great that more people are pursuing a graduate education that has them in school after age 25.”

    Why is that “great”? You just stated that like it was obviously true, and never backed it up, either explicitly or implicitly.
    And then you made the Bryan Caplan argument against that notion…
    (My take is that most graduate education is either purely wasteful, or mostly consumption for the student.)

    In any case, one thing to consider when asking economists whether productivity growth looks low because of measurement problems, is that you’re asking economists whether economists are incompetent. So asking them this probably isn’t all that useful, because their identity prevents them from answering truthfully. (If you ask me, in this area, they are incompetent. Massive measurement problems are endemic to the profession, a fact about which they are all in denial. There’s a reason so many of the measures you reported seem to contradict each other.)

  100. dopefishjustin says:

    I recall seeing somewhere that increasing numbers of men in prison are a potential explanation for the decline in PAMLFPR while official unemployment stays low, since prisoners don’t count in the latter. On the face of it it seems plausible that the US’ unprecedented level of mass incarceration would have some impact. Has anyone looked into this formally?

    • Brad says:

      For the denominator of the LFPR the BLS uses the civilian non-institutionalized population. That excludes those in jail, prison, mental hospitals, nursing homes, etc. as well those on active duty in the military.

  101. Blogospheroid says:

    Looking at the future, there are a couple of things which I think can help.

    More talk of wage subsidy instead of minimum wages can definitely help. Accepting the market rate and filling in only the missing part to make a living wage will reduce the cost of the interventions when compared to UBI.

    A serious missing piece of social technology seems to be skill value prediction markets. We know the skills that people paste on linked in,possibly with certifications as well. Get a system to run sufficiently anonymized queries on the tax (IRS) DB to correlate with that and get what they earn. “people with this skill earned this 10 percentile:median:99 percentile” Put up these numbers regularly to get an idea of what skills are valuable. After a few months or years of publishing these, subsidise prediction markets for certifications and later for all listed skills. Let people get an idea of what skills are expected to be valuable in the next 4 years or so. Anything longer may have really wide ranges. It might also be possible to run region specific markets if the tradeoff of liquidity is not too much.

    It seems weird to me your savings which can be diversified are tracked so nicely and “adequately” and your skills which are super difficult to diversify are not being tracked that well.

  102. Guy in TN says:

    We’ve created an economic system in which automation, and the resulting gains in productivity, could actually decrease the average person’s wealth. This is bonkers.

    As other commentators have pointed out, automation is only a problem due to our distributional framework. That is, if we continue to insist that the non-disabled, non-elderly, non-capital-owners must work to receive wealth. The “those who don’t work, don’t eat” ethic, and its corresponding economic systems, makes sense in a primitive world, but it does not make sense in a world of automation.

    I think that as our economic and technological environment evolves, our culture and value systems will evolve with it. Particularity, things like the “Protestant work ethic” or aversion to “handouts” will be cast aside, as people are born into a world where maintaining such ethics is incompatible.

    We have the ability to choose how the gains of automation are distributed. It’s not set in stone that the %0.01 must own the robots- the distribution of property is determined politically. If people haven’t chosen to wield this democratic power yet, its only because they haven’t been pushed far enough to deem it necessary.

    • sharper13 says:

      We’ve created an economic system in which automation, and the resulting gains in productivity, could actually decrease the average person’s wealth.

      It seems to me that you’re assuming your conclusion here. What’s the best evidence that an increase in automation and productivity would decrease the average person’s wealth? I’m coming from the empirical perspective that it never has in the past and I wouldn’t expect it to in the future. If you think that’s wrong, I’d appreciate some evidence pointing in that direction.

      Thanks.

      • Guy in TN says:

        My comment was intended to describe what would happen if technological automation is as the doom-sayers say it is. I’m not trying to make the case that it is- after reading Scott’s post, I realize that question is above my grade.

        If we get to the point where nearly all of the gains of automation are going to a tiny fraction of society, and 99.99% of the public is jobless, it will only be because we democratically consented to this situation (assuming we maintain a democratic society that is able to enforce the rule of law).

  103. Nate the Albatross says:

    I liked the BLS chart joke. I get jokes. On Germany, my reading indicates Germany has a number of structural and culture barriers to women in the workplace. Like half-day elementary school and the stereotype that working women are infertile. So their high rate of employment of prime age men isn’t as impressive.

    The key point I think is the change in high, medium and low paying jobs. Automating high paying jobs is difficult both technically and legally. It makes more sense to design tech to help doctors than to design tech to replace doctors. Low wage jobs aren’t as lucrative to replace unless it is done at scale. Middle class jobs are more numerous than high paying jobs, and it can be worth it for even a small company to replace a relatively small number of them.

    For a lot of low paying jobs, automation seems to threaten the economic model. Without retail, restaurant and truck driving jobs how will Walmart and McDonald’s have customers? There seems to be strong incentives for each company to automate but automation hurts them as a group.

    There are jobs that are robot resistant: parents, pre-school teachers, dog trainers, fishing guides, artists, musicians, pastors… but I’m skeptical anything other than a universal basic income will solve the participation rate problem.

    • BlindKungFuMaster says:

      If you look at the chart with the Germany joke again, you will notice that Germany is a negative outlier in 1990 and not a positive outlier in 2014. So the joke doesn’t really work and there is no need for an explanation beyond “in 1990 lots of eastern Germans had lost their jobs”.

  104. theredsheep says:

    I work as a pharmacy tech at a hospital. It’s occurred to me, more than once, that there’s nothing to prevent most of my job from being automated using existing technology; you’d just have to design a hospital, and its computer system, to accommodate a bunch of robots. It’d be rather expensive. Given that hospitals are enormous investments and you typically only see a few new ones being built at any given time, I think I’m safe for some time. My hospital is still using retrofitted computerized medicine cabinets from the eighties. They, and probably everyone else without an absolutely gigantic endowment, will delay the costs involved in automation for as long as possible. There might be legal issues and challenges as well; you might have difficulty convincing a judge or review panel that it’s okay to trust machines with the delivery of life-saving (and potentially life-ending) medicines. As noted here, the largest technical barrier is probably simple stuff like climbing stairs and navigating around other workers in the way. The things people find hard–multitasking, processing multiple near-simultaneous requests, knowing your way around the hospital–would be quite simple for a machine. And you could be quite confident that it wouldn’t steal the narcs.

    I’m looking into the local community college’s physical therapy assistant program. Annoying as it is to have a person come around and tell you to stretch your aching leg muscles, it’s got to be downright infuriating to take it from a robot. (Also it pays better, etc.)

    So, uh, there’s my worthless anecdotal evidence.

    • Mark V Anderson says:

      Actually very interesting to read your thoughts. And I think you probably have it right on all points. Physical therapy will be one of the last things to be automated, and in the meantime the demand is likely to increase quickly.

  105. resalisbury says:

    An argument that often gets overlooked is that housing costs in high productivity cities are crowding out potential increases in employment and productivity.

    What if we are as good as ever at fundamental technological aspect of productivity innovations but have gotten worse at moving people to places where those innovations are occurring?

    Interstate labor mobility rates are at an all time low since the 1960s and housing costs in our most productive cities are at an all time high.

    Hsieh and Moretti (2017) argue in “Housing Constrains and Spatial Misallocation” that increases in housing costs due primarily to zoning have reduced aggregate GDP growth by 50% since 1960. That is huge. Even if they are off by an order of magnitude, it could still help explain several pp of the reduction in prime age workforce. Additional workers have to come from somewhere…

    https://faculty.chicagobooth.edu/chang-tai.hsieh/research/growth.pdf

    In San Francisco the median housing price is over $1 million. Not everyone needs to move here and be a programmer there’s lots of other work to be done. Unfortunately, most people can’t afford to live in the city so they sleep in cars or commute in 2 hours just to work. Imagine how many more people would work if they could just live here.

    The Economist argued that if not for housing costs the population of the Bay Area would be around 30 million (can’t find the link) instead of ~7 million. That’s just one region where the employment growth would impact things on a national scale.

    Imagine what would have happened to Detroit if in the early years housing prices spike to over $1 million. Or if in the 1860s when millions were migrating to the United States they arrived to discover that the median housing cost $1 million. They would write home and tell friends not to come.

    Housing costs in the last few decades have finally reached a point where they have become a constraint on growth that was not true during previous periods of economic expansion in the US.

    What to do about housing costs? Obama’s outgoing team of economic advisors published a paper with policy recommendations which can be see here. Basically if calls for increased density and by right zoning (ie you can’t put someone’s bulding permit through 3 years of discretionary review and comment).

    https://www.whitehouse.gov/sites/whitehouse.gov/files/images/Housing_Development_Toolkit%20f.2.pdf

    • Brad says:

      This is the downside to the American fetish for localism. Bay Area incumbent residents are essentially screwing over the entire country. It’s similar in a lot of ways to the situation in England before fee tails were reined in and property could be locked up indefinitely by the dead hand of the past.

      The solution is to move zoning decisions, at least for economically critical zones ™, up to the state or better federal level.

    • herculesorion says:

      It’s not the regulations or zoning, bro. It’s gonna take a hell of a lot of building before anyone can bring SF Bay Area prices down through inventory expansion.

      Like, they’re putting in huge apartment arcologies in the San Jose area, thousands of units, and even those are renting single-bedroom places for $3000 a month. New houses a hundred miles away from the city are selling for $800,000. It’s going to take millions of places before the market price is affected one iota by increased building, and that’s not something that you’re going to fix by just tweaking the regulations a bit. It’s more along the lines of Ready Player One-style container-house stacks.

      • Cliff says:

        Pretty much all economists disagree with you, which should bother you.

        • herculesorion says:

          I’d think any economist would consider it trivially true that absolute and relative are two different things. Nowhere in my post do I disagree that increased numbers of houses will lower prices, but it’s going to take to be a lot more of an increase than anyone currently imagines before the wants driving California housing demand are satisfied.

      • resalisbury says:

        Correct. The state of California needs to build 3.5 million homes by 2025. Legislation can have an impact. For instance SB 827 currently in the state senate could open the way for 1 to 2 million homes by increasing the minimum building height for any place within 0.5 miles of transit.

        What remains to be see. Is whether the legislation passes at the state level and then whether localities through up further obstacles.

        • herculesorion says:

          “could open the way for 1 to 2 million (with the potential for local governments to stop it)” is…not gonna cut the mustard.

    • Worley says:

      You’ve got it backwards — people who live in areas with high economic opportunity (and especially homeowners) have gotten to be very good at maximizing their housing values and shutting out competing workers. (And especially insuring the children of less-skilled people don’t show up in the same schools as their children.)

      Remember, all of this is controlled by local laws, so the definition of “good” is “what is beneficial to the people who already live here“.

  106. Freddie deBoer says:

    The commenters here could stand to learn this lesson: you do not live in important times. There is nothing remotely notable about now from the perspective of history.

    • herculesorion says:

      Well. This time is important to me, because I’m freakin’ living in it. I kind of really do care where things are going, and why, because–as you so smugly point out–I’m not special and the world doesn’t care if it rolls me over and grinds me to powder.

      • Freddie deBoer says:

        Sure. And same. I mean that we don’t live in some sort of historically significant epoch. We’re not at the beginning of anything and we’re not at the ending of anything. So much of the Singularity stuff, the AI threat stuff, even some of the way people talk about global warming, seems to be distorted by people’s need to feel like they live at the beginning/end of history.

        • albatross11 says:

          Freddie:

          Over the last few hundred years, changes in technology and science and ideology and economics have meant that past human experiences were often not great guides for the future. Indeed, it looks to me like a majority of human lifetimes in the last century and a large fraction in the last 2-3 centuries would see huge changes in how they lived and what their world looked like, based on these changes. So I think it’s quite likely that we live in an interesting time, because almost all times are pretty interesting at this point. And I expect that to continue until we somehow wipe ourselves out or hit some kind of fundamental limit to this kind of change.

          As an example, my grandmother grew up in a farmhouse with neither electricity nor indoor plumbing, but lived most of her life in what must have seemed like great luxury by comparison. My boss at work grew up during the Cultural Revolution, and now lives in the US. (And apparently her hometown is unrecognizable because of all the growth since she left.) My wife’s grandmother was born in Austria-Hungary in the last years of its existence, survived WW2, and immigrated here. I grew up under the assumption my whole country (and really most of the northern hemisphere) would be incinerated sooner or later; sometime around college, we as a species seem to have called off that plan. And so on.

        • herculesorion says:

          I’d say that one of the not-special things about this time is that people living in it consider it to be The Most Special Time Ever.

          A certain number of persons have always lived in an apocalyptic fantasy, which is understandable, because even the most boring life is given weight and meaning by being The Last Life.

    • mobile says:

      Corollary: the next election is not the most important election in the history of the country/world

      • Freddie deBoer says:

        Indeed, there’s no reason to believe that Trump will be looked back on as an especially notable president.

        • Mark V Anderson says:

          Oh I think he will be notable. Probably not in a good way, but who knows? But certainly he will be noted as an extreme outlier. Unless future Presidents are as off-the-wall as Trump. Even then, he’d be noted as the first one.

      • The Nybbler says:

        Corollary: the next election is not the most important election in the history of the country/world

        Does not follow. That “now” is not very significant doesn’t mean that 5 years from now won’t be. In fact, it could be that the wrong choice results in a significant near-future whereas the right one doesn’t, in which case the election could be important even if the resulting time is not.

    • The Nybbler says:

      Now is pretty boring. About 30 years ago was a pretty important time. Thirty years from now might be also. Even tomorrow might be, though only if Kim Jong Un gets too restive.

    • Wrong Species says:

      Imagine living during the first industrial revolution. You see all the changes happening around you and based on those changes, propose that a radical reordering of society is bound to take place. I could argue that people have thought this in the past and they were proven wrong. And it would be a good argument, until it wasn’t. I don’t know if we are living in a special time, but you can’t dismiss it out of hand.

    • Cliff says:

      Where did you learn that “lesson”? Because it is exactly wrong. We live in the most important time in the history of the universe. The idea that the present time is not remotely notable from the perspective of history is risible.

      For the first time in the billions of years of the universe’s existence there could be monumental developments such as the singularity. Unless you have really interesting ideas about alien civilizations, it’s hard to imagine any previous time period that could compare. Surely, objectively, humanity is the most interesting thing to happen at least on this planet, with the importance of the times increasing monotonically if not exponentially up to the present.

    • pontifex says:

      The commenters here could stand to learn this lesson: you do not live in important times. There is nothing remotely notable about now from the perspective of history.

      How do you know?

    • Scott Alexander says:

      To a first approximation, only one thing happened in history. It’s reasonable to live in fear of a second thing happening.

    • Swami says:

      “There is nothing remotely notable about now from the perspective of history.”

      There is something extremely notable which has never occurred in all history, though it may stretch our concept of “now”.

      The breakthrough which we have seen for the past two centuries is that it is the only time in history that we have seen widespread long term cumulative improvement in human living conditions. Starting in Britain and the US and expanding slowly out from there, we have seen lifespan, health, freedom, income, wealth, education, tolerance, and equality of opportunity increase every generation. In other words, for the first time in 13.8 billion years we can see clear signs of cumulative long term progress. This isn’t just a first for humans, it is the only time it has ever happened for any known species.

      Now, one could counter that the current era “now” is not especially significant compared to other eras post IR. But even that is wrong. What is special is the rate of progress for the last generation globally. Never before have we seen such rapid improvements in the above factors globally. Poverty, education and equality have improved at a faster pace over the last generation (a billion people emerging out of extreme poverty) than at any time ever.

      Denial of this as a special time just reflects complete denial of the advancement of living conditions and the expansion on the range globally of progress we are witnessing in this age.

  107. stephentrobbins says:

    Technology Change Not the Culprit in Wages Falling Behind US Productivity Gains
    https://www.nakedcapitalism.com/2018/02/technology-change-not-culprit-wages-falling-behind-us-productivity-gains.html

    This seems to be relevant if marginal evidence against “even if robots aren’t putting people out of work, they may be causing wages to stagnate”.

  108. Neutrino says:

    Scott,
    Thank you for a fascinating article.
    Here is a link to a graph to consider for subsequent analysis, perhaps already completed elsewhere.

    https://upload.wikimedia.org/wikipedia/commons/c/c5/IQuoteWeltEngl.PNG

    The graph shows Gross Fixed Capital Formation trends as a percent of GDP for OECD countries and for the world from 1960 through 2012. The downward secular trend is troublesome as it overlaps the latter stages of The Glorious Thirty post-war years and subsequent upheavals to world economic growth.

  109. LepidopteristBB says:

    Hi Scott–longtime fan, first time poster.

    I tremendously enjoyed this entry just as I do the vast majority of your posts. I would love to see you do a follow-up post on a closely related subject to technological unemployment–technological/economic unmarriageability. This is a real problem in 2018, particularly for men, particularly for the African American community.

    Thank you very much.

    • Randy M says:

      What is the technological angle here? Increased production->welfare->reduced need for 2-parent household?

      • LepidopteristBB says:

        1: Women no longer need a male partner for economic security as they have their own careers equal to any man’s in the workforce.
        2: Lower- and lower-middle class men are being pushed by the technological revolution out of careers one can reasonably raise a family on and that is highly unattractive to prospective mates.
        3: With increased artificial insemination, single-parent adoption, and childfree-by-choice living, women have less and less of a need of a man to have children with.
        4: With it being easier to collect assistance as a single parent there is even less demand for a breadwinner in the house.

        I am going to guess that the above phenomenon explains a *lot* of the rise of suicide and opioid abuse in white males, and a lot of the rise of homelessness and criminality in minority communities.

  110. JohnBuridan says:

    I think Scott should have finnessed the article more to factor in IQ and even the Big 5 into his picture. One of the interesting things Autor said was that certain skills will remain bundled with skills that are automatable, so a human can use a robot but will stil be required for those hard-to-automate skills: interpersonal interaction, flexibility, adaptability, and problem solving. I might add though that “problem solving” is an IQ thing, while everything else here is essentially interpersonal/communication skills.

    I find it hard to believe that the missing men, men who already are undereducated, will take jobs in which their primary role is interpersonal in nature. If we had some statistics about the IQ and personality traits of these men, we would have a better idea of what is holding them back. I predict it would be very clear that particular type of person is becoming underemployed or unemployed in our society, and that while previous downturns were temporary, they were also more egalitarian in the types of people they knocked back.

    Presumably these men have higher testosterone and prefer taking risks and doing active labor, being generally weak on interpersonal skills, and having lower IQs, are thus not great problem solvers and not highly adaptable.

    I would predict job polarization to continue increasing and with it more workers will become underemployed, while others join the Missing. (I consider the end of From Dawn to Decadence in which Jacques Barzun paints a picture in which most of society forms an enormous class whose basic needs are met, but also who have nothing to do except maintain their social and cultural networks. The rest of the middle class doesn’t work and not working, democratic republics stop being a dominant form of government, since the middle classes which demanded them in the first place have collapsed into a low productivity and low education equilibrium. Engineers and Raikoth.net rule and are praised for their contribution keeping the machines working.)

    • Molly says:

      Yes, that seems important.

      It looks like the trend is replacing jobs where it was alright to be rather rough around the edges and curse at each other from time to time with jobs that are somewhat socially fragile (whether up, in the case of PC workplaces, or down, in the case of health aids and retail), and it would be useful to know if that’s something that’s actually important, or simply a mirage.

      From what I heard the effects of IQ and conscientiousness have been studied, but I don’t know anything about the effects of the other traits. From observation, many service economy jobs are ones introverted, disagreeable people *can* do, but perhaps not long term, and not nearly so well as extroverted, agreeable people. But there are also a bunch of new shipping jobs, I think? It would be useful to see numbers on that.

  111. Eponymous says:

    This working paper just came out:

    Explaining the Decline in the U.S. Employment-to-Population Ratio: A Review of the Evidence

    This paper first documents trends in employment rates and then reviews what is known about the various factors that have been proposed to explain the decline in the overall employment-to-population ratio between 1999 and 2016. Population aging has had a notable effect on the overall employment rate over this period, but within-age-group declines in employment among young and prime age adults have been at least as important. Our review of the evidence leads us to conclude that labor demand factors, in particular trade and the penetration of robots into the labor market, are the most important drivers of observed within-group declines in employment. Labor supply factors, most notably increased participation in disability insurance programs, have played a less important but not inconsequential role. Increases in the real value of the minimum wage and in the share of individuals with prison records also have contributed modestly to the decline in the aggregate employment rate. In addition to these factors, whose effects we roughly quantify, we also identify a set of potentially important factors about which the evidence is too preliminary to draw any clear conclusion. These include improvements in leisure technology, changing social norms, increased drug use, growth in occupational licensing, and the costs and challenges associated with child care. Our evidence-driven ranking of factors should be useful for guiding future discussions about the sources of decline in the aggregate employment-to-population ratio and consequently the likely efficacy of alternative policy approaches to increasing employment rates.

    http://www.nber.org/papers/w24333

  112. fr8train_ssc says:

    I see a significant omission in the comments from transhumanists about the potential for human augmentation to go hand in hand with automation. That is, if robots can put a dent in the cost disease for medicine, and BCI techniques improve, it may become more cost efficient to have a human in the loop augmenting a robot specialized for the task. I can see this happening especially for military applications.

    I’m also skeptical of a full robotic takeover of labor when we look at fundamental economics of it, though from less of an Austrian school perspective, and more of a material/supply chain perspective. Mining is highly capital intensive, and while there’s still a good amount left to automate, there’s less room to cut costs there compared to farming. If robotics continue to reduce the cost food (which is something I’m involved with) along with new advances in material science for construction (Think Super Wood) then the relative costs of metals and minerals needed for wires/semiconductors and inputs for making robots will increase compared to the relative cost for keeping humans alive.

    In the worst case, we live in a neo-feudal society, except serfdom doesn’t mean being tied to the land, so much as you being tied by your cybernetic enhancements to a corporation. In the best case we live post scarcity, but don’t have to worry about a central organization democratizing robots because we’ll have embedded our own productivity into ourselves.

    But given that we live in the stupidist timeline (not the worst or best) I’ll leave it to SMBC to argue what the future holds (https://www.smbc-comics.com/comic/specialization)

  113. l33tminion says:

    But about 30% of people on disability do say they want a job. It’s unclear exactly what they mean

    A lot of disabilities are intermittent (that’s alluded to in this very post, with the hypothetical factory worker whose back pain “flares up”). I’d guess a lot of people on disabilities want a job with enough flexibility/slack that they don’t get fired if they can only do it some of the time.

  114. Joseph Greenwood says:

    Every time I read this post’s heading, I see “Teleological unemployment: much more than you wanted to know.”

  115. pjiq says:

    Scott: Just graduated with a bachelor’s in economics, and while I am no expert, I have learned this much about the economics profession:

    It’s a LOT of theory.

    Economics is like philosophy with graphs. That’s why I took it. What could be better than philosophy with graphs! And it’s certainly very fun. But it’s also very political, very morally charged, and very much about proving what people want to prove. Even though a lot of statistical data is analyzed, it’s analyzed using the lens of “worker productivity” and “human capital” and “deadweight loss”, terms that imply that rich people deserve their $$ and all regulation is bad. Which might all be true, but when your models guarantee such conclusions, it’s kind of weird that they analyze such massive datasets just to make their arguments more convincing.

    Of course, econometric analysis is still useful, and not everything economists say is wrong. But it’s surprisingly consistent in it’s implications, in my opinion because of the theories it uses to interpret things.

    In my opinion:

    1) you’re right about “there have always been Luddites” historical arguments being very convincing.

    2) you’re right also right that these people are probably wrong, in that at some point it seems logically it would have to have some kind of serious effect.

    3) I agree with Caplan that education is already too big and is not the solution to tech unemployment. Robots aren’t going to automatically start at the bottom of our economic order/ with the least educated jobs: minimum wage jobs first, then $10 an hour jobs, etc. They’ll start wherever they start, with $40,000 a year trucking jobs, with programming jobs, with whatever they figure out how to do first.

    Other thoughts:

    4) expansion of government spending/ jobs that are ultimately funded by government $$ implies that more jobs could be marginally less productive and survive for longer, because of the lack of a short term market feedback mechanism for those jobs’ funding. This might be the government’s way of dealing with technological unemployment without causing a lot of “moral hazard” by saying “some of you guys just don’t have to work anymore.” If this isn’t happening now, I see it as realistic to expect a government to do this in the near future if tech unemployment actually takes off around self-driving trucks or something.

    5) I think it would be nice if the general discussion of things was not about “what happens if everyone isn’t able to sell their labor for 40 hours every week for the rest of eternity?!! Wouldn’t that be the worst thing ever??” I mean, maybe idealizing human freedom would suggest this isn’t the best of all possible worlds.

  116. Anon256 says:

    Isn’t “disability” often de facto defined (both by the people making the decisions on whether you qualify, and more broadly socially) as “inability to do any plausibly available job”? It’s an inherently relative concept. So improved automation/technology might be a direct cause of the increasing prime age male disability rate. Presumably when horses were automated out of work, it started with the fraction of horses that were “too weak/injured to bother keeping alive” increasing until it encompassed almost everything except thoroughbred racehorses and horses kept for aesthetic purposes. Similarly, technological unemployment could eventually lead to the only human jobs being professional athletes and humans employed for aesthetic purposes, and then everyone who isn’t good enough to be an athlete or attractive enough to be a model could be said to be “disabled”.

  117. foggen says:

    My feeling is that automation and offshoring function in more or less the same way when it comes to the US economy. As far as a corporation is concerned, a low-cost customer service robot and a low-cost Indian customer service rep are effectively the same thing. We’ve been riding this wave for so long that it seems normal to chase down the lowest possible third world wages for manufacturing, then use the lowered price of goods to maintain what looks like a reasonably steady cost of living figure. Salaries in certain high-paid jobs are exploding, which looks from the outside like some kind of perverse CEO boy’s club corruption package. I sometimes wonder if what we’re actually seeing is moderate raises for them and an unrelenting series of pay cuts for the rest of us as a kind of hidden inflation wrecks the entire rest of the economy.

  118. microsrfr says:

    Whenever management replaces workers with a machine, they do so because the labor cost to design, manufacture, operate and maintain that machine is less than the cost of the labor replaced. This means that, even with perfect and instantaneous retraining, total wages (and therefore demand) will be reduced. This is particularly true in an age of monopolies where pricing is at best inelastic.

    The Industrial Revolution between 1900 and 1930 saw the cost of manual work reduced about a factor of one thousand. As a result, we saw a Great Depression, a World War and a post-war inflation. Stalin even murdered millions of peasants made excess by farm machines. Over the past 50 years we have seen the cost of computation fall by a factor of one trillion. Already businesses are capitulating on growth and resorting to massive stock buybacks to prop up their stock in the face of shrinking demand. Creating new products will not help when consumers cannot support demand for those products already being produced. This is particularly true since most new companies have extremely small payroll to sales ratios. Where is demand to come from?

    Now we are facing rapid acceleration of automation due to the advent of learning machines to be followed by machines that learn how to learn. I estimate, for example, that self-driving cars will ultimately eliminate 10 million jobs in the US. Global economies and democracies are entering an era of great risk and instability.

  119. po8crg says:

    In re disability, the UK government once introduced an “all-work” test for disability, which in theory required anyone claiming disability to be unable to do any job.

    The immediate reaction was that, obviously, since it was possible to do the job of Lucasian Professor of Mathematics with the severe disability that Stephen Hawking has, then almost no-one would pass it.

    Naturally, the actual definition was rather less strict than that.

    The point here is that any actual disability status has to be based on an interaction between the skills and qualifications people have to work and their physical (or mental) disabilities.

    Many people are not qualified for any job that involves working at a desk. If you do a manual job, then you’re much more likely to pick up a disability that disqualifies you from doing such a job than if you’re doing a desk job, partly because of the higher risks of injury when doing manual labor, and partly because the extent of physical damage required to stop you is lower.

    If mid-level manual jobs (ie highly-skilled for manual labor) are the ones being squeezed most by robots, then heading for disability would make a lot of sense.

  120. Tom Bradley says:

    thanks to this great loan company who offer me a loan amount of $82,000USD i was so confuse till someone directed me to this great company who grant loan with any delay you can contact the company for any kind of loan contact email (tombradinvestment@gmail.com)

  121. MatthewCStein says:

    Thank you for doing this research. It’s frustrating that the conclusions are murky, but it’s hard to predict these kind of massive shifts, or even analyze them without the benefit of further years of hindsight. But this is an issue that needs more serious study, quickly, in my amateur opinion.
    I love that you write these in-depth posts, but I wish they could be made more digestible for a wider audience. Would you be ok with me trying to produce a YouTube series on some of your posts like these? I know it’s not the ideal format for the type of detailed citations you provide, but I think I could get more people to sit through a shortish video series than read this post. Considering you usually have charts and graphs involved, I would likely stick those up instead of walls of text. I have rudimentary video editing skills via OpenShot, and I’ve been told I have a nice radio voice.
    (Also, yeah, probably not gonna tackle Moloch any time soon. Hubris of lifting that damn near broke my back, metaphorically. I’m glad I got a recording of the Ginsberg poem then though.)

  122. Worley says:

    In regard to the particular item of the current stagnation of median wages, I have two observations. The first is fairly narrow, and I’ve taken it from elsewhere. The second is more abstract and I’ve not seen it elsewhere.

    It appears that the current set of jobs that are being hit by automation are middle-wage. The commonly-quoted example is managing a fast-food outlet — it used to be the manager had to reorder supplies based on predicted demand, but a lot of that work has been automated by cheap computers. Naturally, this drives down the median wage. Even if the wages of lower-paid jobs (“in-person service”) and higher-paid (“professional”) are both rising, which they are, the median is only sensitive to the group of jobs in the middle of the distribution.

    This particular situation shows signs of putting the job market out of equilibrium. There seems to be a lot of evidence that the distribution of incomes tends to be log-normal — the logs of incomes are normally distributed. And it seems that the income distribution of any demographic or occupational slice tends to be log-normal. A log-normal distribution is produced by taking the product of a large number of random factors, which may suggest the sort of equilibrium processes that cause the income distribution. Log-normal distributions are characterized by only two parameters — one describing the median income and one describing the spread around the median. But a consequence is that you cannot have the pattern we’re seeing now — if the middle and the top grow farther apart, then the middle and the bottom must grow farther apart. So it seems very like that the current situation is deviating from the equilibrium pattern of some powerful process. Of course, that’s not particularly interesting politically, since the income distribution may settle down into a form where the median at equilibrium is lower, or the spread is much higher. But in some way, there is another shoe left to drop.