A rising tide lifts all boats – 水涨船高
It is a centuries old Chinese proverb; a phrase usually spoken in an optimistic tone. Though for a country plagued by floods, perhaps we can learn more by taking note of what happens to everything not built to survive high water.
The phrase is also used often in politics to describe the general betterment as the overall economy grows and expands. But while technological advancement has brought previously unimaginable wealth and well being to millions, many see the current trends in automation as a worrying forecast of what is yet to come.
To others, this is no more of less different than when the automobile put carriage drivers out of business, or similarly the computer to the typist.
In this post, I’m going to lay out a case that cautions us to hesitate, and consider the future structural transitions of our economy, as automation causes many wages to stagnate, and many more jobs to outright disappear. I do not expect you to agree with me on the prescriptions, but on the prognosis, I hope you are willing to hear out my case.
In today’s political climate, all too few are. In fact, when it comes to addressing concerns regarding automation, only one candidate sticks out.
His name is Andrew Yang, and while only polling around 1% (as of June 7, 2019), his campaign is a grassroots movements more so than any other Democratic candidate in the field.
Despite this, a cursory search of google for terms such as “universal basic income” or “automation candidate” and Andrew Yang is at the top of the search results.
That’s because the primary driver of his campaign is the plan to impose a UBI of 1000$ a month funded through a VAT tax, in order to counteract stagnating real wages and job disenfranchisement due to automation.
As a quick point of note: I am not here to force you to agree to Yang’s policies for 2020. You can call for massive deregulation, you can call for the socialization of all industries, you can call for whatever — I personally would prefer a land value tax over a VAT — but that is not what this post is about. What is important about Yang’s candidacy, is that it’s drawing light to a problem, which the American people must face in the coming decades.
Automation in the 21st Century
The threat of automation is a fairly simple case to make, that all economists can recognize.
As machines increase in productivity, the relative cost of human labor increases. As firms seek to maximize revenue, they must minimize costs, including labor inputs. Wages are decreased, or jobs are replaced. As jobs are replaced, the labor force expands in other industries due to displaced workers, and the competition drives down wages. Overall there is a greater productive capacity, and new jobs can be created in the wake of these gains, perhaps to an even greater extent than there were before — although as we’ll see, this is not always the case.
The Effect on Consumers
Before this, however, there are also interesting effects on the prices of goods and services in an economy I want to look at first. You may have seen charts such as the following circulating your twitter or facebook feed:
This is known as Baumol’s Cost Disease, and it helps to explain the heavy differentiation in the price of goods. Cheap mass produceable goods have become so due to automation, while comparatively many more specialized jobs now require higher skillsets, or advanced manufacturing that necessitates high skill labor inputs.
Please note that healthcare, housing and education have many other factors causing their rising cost to the general public, and that each one of them could warrant a blog post to the merit of their own complexity — in fact I have already done so with healthcare.
But these issues — healthcare, housing and education, are at the core of our domestic policy, and will be discussed frequently in the debates. It is important to know that automation does have a role in the exacerbation of costs.
The Effect on Labor
In previous generations, economic expansion occurred in a differently structured economy than the one we live in today. MIT Economist Daron Acemoglu illustrates through empirical data how automation is affecting much our labor force in a new paper here (free connecting data also by Acemoglu here).
Compared to prior generations, the displacement effect of automation is fairly similar. Just as when the automobile replaced the carriage, laborers were left in structural unemployment, with a need for a new skillset, education or training. Looking at our modern era, automation is more prone to creating jobs in which human labor inputs are relatively more expensive, and it is occurring at a higher velocity than in previous generations.
When combined with a labor force in the 21st century that is increasingly demanding better skillsets and/or specialization over time, this creates a milieu in which displaced workers are finding it harder to enter the labor force for existing industries, which is dangerous when doubled with the fact that new jobs created by automation are also becoming less likely to be accommodating to low-skill human labor inputs.
The combination of these two factors is quite troubling, and has affected many millions of Americans. Many attribute President Donald Trump’s victory in 2016 as an appeal to such displaced people in the rust belt, promising them that they would not be left behind under his economy. I will speak more on this in a moment, but one point this is accurate; manufacturing job loss within the United States is definitely a part of the equation.
However, while President Trump has attributed much of this decline to China, there seems to be a larger long term issue domestically.
Data from various sources concerning domestic worker compensation shows an increasing “productivity gap.” This shows that while outsourcing surely does have an effect on our domestic productivity, there is an internal problem affecting workers as well. In fact an over-focus on China specifically for outsourcing is hardly warranted: their demographic dividend of non-dependent youth is quickly drying up, and despite their “Made in China 2025” initiative, manufacturing is likely moving elsewhere (1 & 2) as the shifting structure of their economy necessitates the expansion of a service sector.
But I digress, as I have a bias in my studies to focus on the Chinese economy.
I’ve talked with a lot of my friends about the “Productivity Gap,” and have most prominently referred to this great post trying to go into the mechanics of what is going on.
After reviewing a lot of the base data that is put into these measurements, I think I agree with the assertion that the productivity gap is being over-classified, and truly is reflecting a split in relative share of workers’ wages from revenue. I am also likely going to try to make my own post an establish my own productivity ratings in the primary, secondary and tertiary sectors, but such an undertaking will likely take a very long time.
While currently employment is doing well in a fairly strong economy (June 7th, 2019, so if we crash tomorrow I can save face), we are beginning to see the first effects of the coming tide of automation.
Note that Labor Force Participation (BLS statistics) can explain why we’re not yet seeing a UE effect from automation, but only since around 2000, and there are many other mitigating factors to consider – population demographics, minimum wage law, general welfare policy, etc.
In response to a recent article by Economics21.org, there is an interesting contention that it’s actually our slowdown in automation that has led our productivity to drop, and thus our profitability and willingness to hire more labor, that has accelerated job loss.
Of course, job displacement is not the problem as I cited earlier: it is job reinstatement in our new economy that is most concerning.
But for more data, here is a paper out of the Brookings institute that shows you can’t blame capital deepening (or lack thereof) on the recent slowdown in productivity.
Another recent paper published just a year ago looked at our economy and automation, and came to a fairly traditional conclusion: automation has increased our overall wealth. Along with population growth, the total number of jobs has increased as well.
Unfortunately, the paper also shows that there has been a displacement of various sects of our labor force already, and this has caused downward pressure to be put on wages, as the supply of labor in existing industries is now being expanded upon.
This is what I spoke of earlier on; we are at the early stages of a wave of automation the likes of which appear to be different than any prior shift. AI and computerization are going to affect our economy in ways few, if any, can fathom. It will have an effect on consumers’ ability to obtain healthcare, education, and housing, and laborers’ overall well being.
For now, the topic is too far from the mainstream; we are busy dealing with a seemingly pointless trade war centered around tariff regimes that don’t seem to be effectively reaching any economic or political goal.
On neither side of the aisle do we see this debated even close to the degree it merits.
Automation is not necessarily a bad thing, in fact it is my personal view that the technological advancement of society is mankind’s greatest feature. But the political and economic systems we currently have in place may not be a strong enough levee to survive our own ambition. If we don’t think to the future, it’s hard to say if our economy will sink or survive in the rising tides.
*UPDATE 08.11.2019 EUR*
I’ve seen more and more debates unfold over the last few months surrounding automation, and sadly it feels as if the two main groups are both making… incomplete arguments.
The first group are those concerned with automation, but present it zealously, claiming it has killed millions of jobs. This is incorrect; the predictions on net job loss are future projections, right now all we have seen is job transformation.
But then we come to the second group, which likes to only address the weaker arguments, and ignores broader points. @Noahpinion on twitter demonstrates this in a long thread where he spells out just how automation isn’t a problem. During these exchanges, he encountered one such Yang fan who believed that millions of people were out of jobs in certain sectors… Noah showed the FRED data proving him wrong but failed to do any sort of follow up.
Dig a little bit deeper, and you’ll find that BLS Labor predictions show telephone operators and cashiers indeed decreasing in employment number *and* percent over the coming decade.
If nothing else, I hope this post helps to frame the debate in a correct light: automation has not already destroyed millions of lives, but it will be having negative effects on workers over the coming decades, in a fashion we have not previously experienced, and should prepare for.
*END OF UPDATE*