Sunday, February 23, 2020

Andrew Yang and Automation

Rome fell in 476 AD, according to the high school and World History 101 shorthand we are used to. You can prefer a date in the 370's instead, or 410, or the Vandals taking Carthage in 439 AD, which broke the tax spine of the Western Empire for good.  Going in the other direction, you can choose the collapse of the reconquest by Justinian in the late 6th C, or even later.  If one wants to be really technical, Constantinople, the Eastern capital of the Roman Empire for a thousand years after Rome itself fell to Ostrogoths, did not fall until 1453. Few would pick that date, but you could, and make an argument with at least some facts to buttress it.

But let's focus on 476 and let that hover in the background as we look at the collapse.  In 440AD all looked bleak for the empire, though those in the central cities did not perceive it.  Britain and Northern Gaul had fallen out of the empire. Northern Africa was a trading partner, not part of the empire. The influence over Persia, Syria, and the easternmost sections of empire was waning. Yet by 450, trading was bustling again.  Some researchers would claim that this was actually the height of trade throughout the Mediterranean, unmatched for more than a thousand years. There was recovery! Despite all the dark portents, Rome reached its height.  Arguably.  Some would pick 150, 350, or other dates. Still, a case could be made.

If you were living at the time, those naysayers who pointed to the loss of tax revenue from across the Mediterranean, or the growing power of the Goths, who had an internal kingdom in Gaul, or the slow loss of border provinces and increase in cross-border raiding would be laughed off.  However plausible their arguments might sound that the empire was in decline, the objective evidence was that things were fine.

Just ten years later. 440AD to 450AD. They didn't know it, but the collapse was indeed near, because long-term secular trends were overwhelming short-term local ones.

I think of this when I read Andrew Yang's worries about the coming jobs catastrophe. He believes that automation is going to destroy more jobs than it creates.  Note that he is not a typical doomsayer.  He does understand that jobs will be created as well as destroyed.  He thinks the latter will be greater.* Truck drivers and cashiers are on the near horizon of disappearing jobs. There are millions of those.  The jobs created in a new economy will have to be very impressive to keep up with that over the next three decades.

He was honest with Democrats that job retraining mostly doesn't work.  Republicans didn't notice this courage, Democrats generally just whistled past the graveyard.

The traditional argument against this is powerful.  I have seen Texan99 make it several times. Despite the impression we always have on this side of the divide that more jobs will be lost, the consistent experience of humankind is that there are actually more jobs when we get there.  There are fewer jobs for buggy-whip makers and manure-shovelers, but lots and lots of automotive jobs. That might be the way to bet.  However, I am reminded of Nicholas Nassim Taleb's description of the Lebanese who believed that after a thousand years of Christians and Muslims living together, the 20th C difficulties would end the same way.  So they moved to Cyprus decades ago to wait out the temporary antagonism, staying in hotels there.

They are still in hotels in Cyprus.

I don't take on faith that there are other jobs that will show up as Amazon automates delivery, both in the warehouse and on the highways.  I'd like to think there are other jobs for the low-skilled working at $ in convenience stores.  Such jobs have appeared before. Yet I don't see any guarantee of that. The past is the best predictor of the future, but it is not an infallible predictor of the future.

Things look great at the moment, and in the short term I think that can hold.  Will it hold two more decades?  Four?  What will we do with decent people who would absolutely show up on time and work hard, but there are now ten of them for every four jobs?

*Relatedly, when he touts Universal Basic Income, he does not pretend that there is no risk and no loss of productivity, as most hemp-headed advocates do.  He believes that it is a 70-30, or perhaps only 60-40 tradeoff to give more freedom to the poor. If you dig deeper, he believes that some percentage (I conclude from his comments about 50%) of the poor are generally hardworking Americans, who don't have a lot of skills or whose industries collapsed or had bad luck, who would do fine with a minimal safety net. Those, he thinks will muddle through with just a little help.  He hints that there may be a lot of minorities in this group, who would be better served by supplement rather than rescue, and would generally not take advantage.


David Foster said...

In the actual productivity numbers, though, we're just not seeing the kind of step change that would be expected if robotics/AI were really having the kind of impact advertised by many.

I understand the argument that the big impact hasn't shown up *yet* but is in the pipeline, or just over the horizon...but consider what's already been broadly implemented: there are self-checkout stations in stores of all kinds, and they are being used. They are also showing up in restaurants. Parking garages now rarely have a separate cashier in each lane, it is now automated, with one person supervising multiple lanes (sometimes on different sites). Mail-order transactions that once required talking to a human are now done directly on the Internet. All this and lots more, and productivity growth is still lame.

If you consider the impact of past waves of automation...textile machinery circa 1800...railroads replacing thousands of men with mules and carts...steamships replacing sailing ships and then oil-burners replacing the need for human stokers...tractors replacing mules on the farm...mainframe computers replacing rooms full of human's hard to believe that what is going on now will be so much overwhelmingly greater in its impact.

I think the problem is most likely not that labor productivity increase is too *high*, but rather that it is too *low*. Higher productivity is the only way to have some chance of digging ourselves out of debt...on-the-books debt and off-the-books debt...while generally improving standards of living.

james said...

There's not just AI to consider, but off-shoring. Combining these with the usual industry fluctuations, the pain in this or that area may grow too great for the usual social web to handle.

sykes.1 said...

The focus on automation is wrong-headed. Of course automation reduces employment, that is Econ 101. The problems our working class have come from excess labor availability, which drives down wages. And it has, working class wages have fallen steadily for at least 40 years. 1965 was probably the peak year for the working class.

The causes are, in no particular order: (1) unskilled women entering the labor pool in very large numbers; (2) legal and, especially, illegal immigration; (3) free trade and the importation of foreign made goods; (4) related to (3) off-shoring of American industry; (5) automation; (6) systematic destruction of private industry unions.

The first five items increase labor supply, and the increased amount of labor reduces its price. The sixth item is more political, but the loss of effective unions leaves workers naked to the predatory practices of businesses.

Automation and women's labor might have been acceptable if political leaders had cutoff the other sources of labor, but instead they reinforced them. They are still pushing items (1) through (4).

What we need is highly restricted (ideally zero) immigration and protectionist trade policies. Those will drive up costs, but they will also eliminate many of the problems of the working class.

Like opioids.

Christopher B said...

A couple of thoughts.

One is an underlying but unstated assumption in the analysis is population in general will continue to increase. This may turn out to be an incorrect assumption in the medium term. We don't see the issue in the U.S because we have the Millenial baby boomlet from 1980-2000ish and because we live next door to a country with a young demography, Mexico. In most of the rest of the world, particularly Europe and Asia but also South America (Africa I'm not sure about), is in the beginning stages of demographic collapse with their native populations dropping below replacement rate. China is the big one as the one-child policy comes back to bite them with a vengeance since they not only have a declining population but a smaller number of mothers that will further reduce their ability to even sustain a stable population.

Japan is quite far down this road, and it's not accident they are a leader in robotics and automation because they've been trying to run a modern industrial economy with a rapidly declining workforce. It maybe that what doesn't show up in the future is the workforce, not the jobs.

The second thought is something along the order of sykes comment. A number of the automation practices that David pointed out (self service gas pumps and check out lines, self service online ordering) are really just drafting customers to be unpaid labor in retail outlets. That undoubtedly makes sense from an individual firm perspective, especially given the explosive growth in mandated benefits and taxes over the last century, but do eliminating those jobs really improve productivity overall? Instead of spending 5 minutes getting my gas tank filled (along with a window wash and oil check) I now spending 10 minutes of my much higher paid time pumping gas. Instead of a short drive to a well stocked hardware store with knowledgeable staff to pick up something to perform a minor home repair, I spending half an hour researching on YouTube, make a longer drive to a big box home improvement store, wander around trying to find the items I need (or can make do with), and then go through their self-service checkout because they are short staffed. Have we really gained that much through the elimination of what could be moderately paid entry level positions? Might those positions make a comeback if we did implement Yang's proposal, especially if we coupled it with some of the other changes that Friedman included when it was first advanced, like the elimination of targeted subsidy programs and maybe a roll back or elimination of minimum wage?

David Foster said...

Christopher B..."A number of the automation practices that David pointed out (self service gas pumps and check out lines, self service online ordering) are really just drafting customers to be unpaid labor in retail outlets. That undoubtedly makes sense from an individual firm perspective, especially given the explosive growth in mandated benefits and taxes over the last century, but do eliminating those jobs really improve productivity overall? Instead of spending 5 minutes getting my gas tank filled (along with a window wash and oil check) I now spending 10 minutes of my much higher paid time pumping gas. "

I think this varies from process to process. At the grocery store, I find it definitely faster (when I have a lot of stuff) to have a professional checker do it; he is much faster and also I don't have to listen to the annoying squawking of the poorly-designed self-check-out system. That's IF the line at the traditional checkout counter is of reasonable size. At a gas station, though, I can pump gas as fast as the gas station guy can, though not much fun in very cold weather.

Some banks now have apps that you can use to deposit checks without going in to the bank. Saves you the drive and the wait in line...the amount of time taken depends largely on the quality of the app.

dmoelling said...

The world hasn't changed that much since the industrial revolution. One consistent issue is that it is difficult to get reliable workers. Henry Ford's $5/day wage was intended to get sober and reliable workers since the assembly line needed a full crew. He also went overboard in checking on the worker's sobriety, but before prohibition it was common for guys to drink away their pay before they got home.

In the 1970's I worked summers at a furniture machinery importer in Michigan. They offered me a job at more than I would get when I graduated with my engineering degree! The reason was that they found it impossible to get guys without drug or booze issues to work independently installing and servicing machines. This was also the case that guys could easily get high pay if they worked for the big 3 auto plants.

The problem of employing ex-cons, addicts, and those with other issues will not change unless our means of resolving these issues improve.

Grim said...

I likewise think we are near a watershed on employment; you’ve read Texan99’s rebuttal at least a few times at my place, where she’s tried to convince me that I am wrong. I may well be, but I don’t think I am. I think we are close to an economy that just doesn’t need most of humanity, except possibly as consumers; and who, though they would gladly work, will be uneconomical to employ. So if they’re going to consume, or even survive, we need a whole new model for resource distribution.

It’s a bigger change than I think many appreciate. It’s the falsification of a Bible verse: “If a man will not work, neither shall he eat.” Practically we’ve fallen away from that for a generation, since LBJ here and longer elsewhere. In principle, though, America has held that work is normative and proper. Welfare to work has always been popular here; handouts are not.

The alternative, though, is to declare that it is fine if people just starve if no one needs or wants them enough to pay them enough to survive. The black market will sustain some of them, since there will always be things people in power want to prohibit. But pushing a lot more people into the black market has consequences too. More police, more prison guards; well, that will provide jobs too. But none of us should want for the solution to look like more crime and more prisons.

Theoretically we could all be artists and creatives, supported by an automated economy that provides us with whatever we really need. Those who can still manage to be useful economically could become fantastically rich, able to pursue dreams like Elon Musk is doing already. It doesn’t have to be ugly, but it’s a huge change we need to think through.

Texan99 said...

I can never quite get past Milton Friedman's parable of the Chinese construction workers with shovels. Asked why they aren't provided with bulldozers, they explain the probable loss of jobs. Friedman asks, "Why not take away their shovels and issue them spoons?"

The choice always looks different if you contemplate taking away the automation and enforcing a return to human labor, than if you imagine the change from human labor to automation. I think that's largely because we have a hard time imagining what form the wider adjustment will take.

Assistant Village Idiot said...

Yes, we have a hard time imagining it, which biases us against it. True.

That is not a proof that it is always going to occur forever.

Texan99 said...

Nothing will ever prove a permanent trend in the future. But we can be on our guard against a firm belief in a future trend if it's built on faulty assumptions and demonstrable bias errors. Sometimes all we can do is go by what's usually happened in the past, absent a convincing argument for a dramatic change, and then do our best to hedge against the unexpected.

David Foster said...

I had a lawn sprinkler system installed last summer...several guys digging shallow trenches, laying pipe, connecting a backflow preventer, installing a control panel, etc. Could have used some automation, in the form of a lightweight trench-digger to replace the shovels, but it's hard to imagine that the whole process is going to be automated.

On Sunday, I was at an airport that is colocated with a large helicopter maintenance base--I believe they are a subcontractor to Boeing. Inspecting parts to see what needs to be replaced, checking records to see what parts have exceeded their hour or cycle limit, stripping paint and repainting, testing the renovated helicopter and certifying that it meets standards...are these tasks all going to be automated?

Talking with someone who has a baby 6 months old...both parents work outside the home, they have nannies but also do a lot of the childcare themselves. Will the nanny function be automated?

Currently involved in a situation involving a building which may (or may not) need major structural repairs. Will the work of the structural engineers who are involved in this situation be automated?

I know a couple of 20-somethings who support themselves as musicians/arrangers. Their jobs were already automated, back in the 1920s, via phonograph, broadcast radio, and sound films...yet somehow, these jobs still exist.

I know several startup companies pretty well. These companies do not just employ "technology geniuses", they need salespeople, deal structurers and dealmakers, etc.

I think that looking at actual jobs done by actual people is a corrective to excessive automatic fears.

Texan99 said...

I used to make a good living as a typist, an obsolete skill no one pays for any more. I was also an ace at re-glazing old wooden windows.

Back then there was no possibility whatever of my telecommuting as a lawyer, running an internet etail business, or for that matter winning or performing a county government office relying almost entirely on social media. The people who implanted and monitor my pacemaker for a price within my means were barely a gleam in anyone's eye; ditto my cataract surgery team. My dentist employs an office full of people who do astonishing work for a reasonable price on root canals, painlessly, within an hour. Each of these people employs countless others for tasks they can afford to pay for only because of jobs that became possible because of automation: that's everything from grocery clerks, nail salon operators, and yard workers to doctors and lawyers. And remember that the businesses that design, build, and install the automating equipment also employ everything from janitors to engineers.

There's a reason an amazing fraction of the world's population has been lifted out of poverty in the last 50 years. It's partly the general efficiency achieved by free markets, but a big part of it is technology and innovative automation. Change is painful and disruptive, especially when it's rapid, but that doesn't automatically make it more destructive than stagnant poverty. And our duty of charity toward anyone who is caught short and unable to change quickly will never go away, with or without automation. Better to help (individually, from our own pockets) than to muffle change, which you can't do, anyway.

David Foster said...

Oh, typing isn't an obsolete skill these day at all, it's just an *uncompensated* skill which must be done in order to exercise other skills which *are* compensated.

I wonder if there are other skills like that...once paid for and done by specialists, but now required as part of the entry to do other things.

Assistant Village Idiot said...

Writing was once that, a specialist's trade that became a universal for any educated person. Driving became that in many fields, including much of sales and consulting. That is waning but still present.

Texan99 said...

Right, typing is like having a driver's license. The lack of it may inhibit your ability to be hired, but having it won't move you up on any lists. Anyway,. the point is that the world changes around you, and your job skills don't confer a lifelong advantage, no matter what Congress might prefer. We can't render people employable by fiat. They're employable if they offer something that other people are willing to pay for.