A friend and I have been going back and forth on automation and it's potential effect on the labour force for a while now. He has won me over to the viewpoint that CURRENTLY the labour market is depressed in western nations due to globalization's effect on labour markets, as well as demographic trends, and not automation. But I still maintain that it's possible in the near future (he doesn't think it's impossible either, but feels that's too speculative to believe strongly in the possibility).
Anyway, the Atlantic decided to do an interesting article on this very thing: A World Without Work
Mostly it's just playing with the possibilities of what might happen if increasing automation brings permanent structural elevated unemployment. Basically, armchair guesswork, with a bit of optimism.
I focused most on section two, where the supporting arguments for a wave of automation are. The horse example is cute, but also genuinely illustrative that permanent obsolescence is possible, we just didn't care because employment for horses isn't exactly an overriding social concern for humans.
Anyway, some of the souring is poor, or forced. He caught this:
NJ wrote:I looked up the Loukas Karabarbounis and Brent Neiman work, and I think he somewhat misrepresents what their paper actually says-- it attributes that fall in labor share to the relative cheapness of investment (http://faculty.chicagobooth.edu/brent.n ... rch/KN.pdf). You have to really torture that paper to get what the author says out of it:But Loukas Karabarbounis and Brent Neiman, economists at the University of Chicago, have estimated that almost half of the decline is the result of businesses’ replacing workers with computers and software.
Well, no. What the paper says is that when investments are cheaper than labor, companies spend more on investments and less on labor. Presumably if labor is more expensive, that goes the other direction. This runs into the same problem as all of the labor pool arguments: you're looking at a period that is coincident with the emergence of women in the workforce and that is coincident with globalization, so there's a massively expanding labor pool available to a given company. Plus, we've been short on major wars for a long while, so there's no massive pullback in labor force that might drive up prices.
Re: the plight of the young college grad: https://en.wikipedia.org/wiki/Education ... ted_States
what would you expect, given the rising incidence of college graduation? We've got nearly 35% BAs and above among 25-29 year-olds-- did we imagine that the entry-level job sector needed 35% college grads?
But one of the other sources is much more solid, as well as an interesting read in it's own right: Oxford study, where they project that machines may be able to perform as much as half of all US jobs within the next 20 years
They go into quite a lot of detail, including some surprisingly in-depth historical research. Crucially, they are not stating that mass automation will happen, merely that it is certainly possible.
The second article is quite long, and is academic (though is largely readable other than the math-y bits and even those are probably perfectly readable if you have some basic stats/econ education), so I just noted some interesting points:
- Technological advances favouring more skilled workers is almost entirely a twentieth century phenomenon. Previous mass waves of technological change actually favoured "deskilling", which is to say they resulted in greater employment for wider numbers of people while requiring fewer skills.
- An important evolutionary step that allowed growth during the later stages of the industrial revolution was the transportation revolution which vastly increased market sizes. Our current economy is unlikely to to feature such a jump in distribution capabilities, given we already have the internet and globalization.
- Polarization of the labour market due to technological advances has multiple historical precedents.
- The labour market has historically bounced back from technological changes by becoming more educated on an overall basis, but in the modern era we are for the first time seeing that this is not having the effect it previously did and there is in fact declining demand for skilled labour in terms of overall quantity - the education and labour markets are starting to decouple.
- The expansion of automation to tasks which might be described as "non-routine" is unprecedented, and largely thanks to an explosion in efforts to break down non-routine tasks into well-defined problems.
- They go very exhaustively into complex tasks and, critical and non-critical, which are already seeing automation. Including this potentially terrifying passage:
Baxter, a 22,000 USD general-purpose robot, provides a well-known example. The robot features an LCD display screen displaying a pair of eyes that take on different expressions depending on the situation. When the robot is first installed or needs to learn a new pattern, no programming is required. A human worker simply guides the robot arms through the motions that will be needed for the task. Baxter then memorises these patterns and can communicate that it has understood its new instructions. While the physical flexibility of Baxter is limited to performing simple operations such as picking up objects and moving them, different standard attachments can be installed on its arms, allowing Baxter to perform a relatively broad scope of manual tasks at low cost (MGI, 2013).
You know the trope of the human worker who has to teach the low-wage temp or foreign worker how to do his job before he's fired, right? Well, we can experience that with robots now!
- They go substantially into detail about task breakdowns to try and ascertain what jobs may be susceptible to automation. There are some big assumptions made, but they're not unsupported.
- For their numerical claims of future potential effect on the job market, some of the math is based on partially subjective premises (given our research and what we know, is this job automatable?). They've lampshaded this seem to have taken fair measures to ensure the subjective portions were as well-based as possible - they sought consensus opinions and also worked from a hard variable list to decide the likelihood of complex task automation.
- They argue that that labour market polarization will be truncated in the coming decades and average wage levels will thereby rise... but only because computerization will fall more heavily on the low-wage jobs, thus eliminating them and leaving higher unemployment rates overall.
One BIG thing I really wish he had more detail or a study link to was the comment in the Atlantic article about new industries being highly labour-efficient. That's massively important for future employment growth and fundamentally crucial to the old idea that new industries bring new jobs as part of the "creative destruction" of industries and labour markets.