Economics

Two Myths About Automation

Dec 12, 2017 Barry Eichengreen

While many people believe that technological progress and job destruction are accelerating dramatically, there is no evidence of either trend. In reality, total factor productivity, the best summary measure of the pace of technical change, has been stagnating since 2005 in the US and across the advanced-country world.

BERKELEY – Robots, machine learning, and artificial intelligence promise to change fundamentally the nature of work. Everyone knows this. Or at least they think they do.

Specifically, they think they know two things. First, more jobs than ever are threatened. “Forrester Predicts that AI-enabled Automation will Eliminate 9% of US Jobs in 2018,” declares one headline. “McKinsey: One-third of US workers could be jobless by 2030 due to automation,” seconds another.

Reports like these leave the impression that technological progress and job destruction are accelerating dramatically. But there is no evidence of either trend. In reality, total factor productivity, the best summary measure of the pace of technical change, has been stagnating since 2005 in the United States and across the advanced-country world.

To continue reading, register now.

Subscribe now for unlimited access to everything PS has to offer.

As a registered user, you can enjoy more PS content every month – for free.

Already have an account? Log in

PS Events: What Economics is Missing

Our latest event, What Economics is Missing, is now live.

Click the link below for opening remarks from Dani Rodrik, followed by a discussion among Ashwini Deshpande, Raquel Fernández, Minouche Shafik, and Vera Songwe on how to achieve inclusivity in economics.

Watch Now

new comment has been posted. new comments have been posted.

0 Comments on this paragraph, 27 Comments on this article

Bob Aceti Dec 31, 2017

"So here’s an idea. Instead of a “tax reform” that allows firms to expense their capital outlays immediately, why not give companies tax credits for the cost of providing lifelong learning to their employees?"
---------------------------------------------------
Good idea. The challenge, currently, is that any 'life-long learning' ("3L") tax credit received would reduce the salary and wages expense. For example, a payroll expense of $10 million plus 3L of $1 Million would result in taxable income of:

$10 M + $1 M - 3L costs (say) $1 M = $10 M = no change

Of course, tax reform may incent companies to invest in employees skills development and training, including unincorporated entities that employ the majority of employees, but the issue of lost productivity of those involved in training for new skills would entail two costs - actual costs for skills training and temporal productivity loss and flight loss - the loss of an employee trained with marketable skills that attract higher compensation from other employer prospects.

We can construct the policy, methods and tax reform required to improve productivity to mitigate lost producvity and skills development costs, without a net benefit to the employer that incurs expenses for training the application of employee 3L tax credits would be less under current tax reform than on status quo tax rules. How so? If you reduce tax rates, each dollar of expense would receive less tax refund or return (yield) than under a higher tax rate.

I think a better idea would be to tax on the basis of revenue by industry sector - i.e.) processing, manufacturing and assembly versus services and banking industries. Apply 3L as a tax-free subsidy to allow employers free money to develop new employee skills. By eliminating the taxable income based on revenue less expenses we would shift business decision-making from tax chicanery - global tax planning - to comprehensive financial analysis that eliminates the tax uncertainty and loop-hole induced system that leaks government tax revenues and reduces the ability of government to engage in directed programs to improve productivity - like the 3L program propsoed.

I am certain that tax authorities will finally recognize that ear-marked tax reforms lead to unnecessary complexity and self-interested politicians that place their electoral funding challenges before the best interests of the global economy. The only unknown is how long it will take before true tax reform, substantially reducing tax complexity driven by Washington tax lobbyists, is placed on the legislative agenda. And that will require reintroduction of an estate tax to curtail the continuing accumulation of significant wealth based on debious criteria: several instances of inheritances that hold major corporate control positions flowing to progeny unworthy and unprepared to deal with the responsibilities of intergenerational wealth succession and increasingly higher-than-average investment growth rates (2014 - Capital in the Twenty-First Century, Thomas Piketty). Distinguished wealth should be effectively taxed at 50% and transfers to offshore trusts and "charitable trusts" should be prohibited. The tax rules are designed by greedy people to prolong a system of privilege similar to the Right of Kings and their Courts that eventually reduced the Brittish Empire to an Island state unable to reconcile its past glories to its common presence - Brexit is a path to further marginalization and isolation that can only be over-come by reducing taxes and chasing the lowest common denominator Dickensian society.
A new reply to this comment has been posted.
manuel werner Dec 20, 2017

The otherwise fine article unfortunately ignores the distributional consequences of robots picking up an increasing share of labour's duties. The health care worker assisted by a robot ends up with a lower share of the fixed resources going to care. This outcome is reflected in the growing wealth and income inequality in western economies, particularly in America. Such imbalances have never ended well (the Roman grain dole was he result of such labour-capital (slaves) substitution and a contributor to the fall of the republic).
- Paul Friesen Dec 20, 2017
  
  Interesting point about slave labour. I really think that is the best model for what is coming. Humans with artificially intelligent machines will be like slave-owning societies of the past.
  
  There are, of course, differences. We really do not need to worry about the rights of machines. They will be intelligent but not self-aware and will not have all the evolved human emotions like anger or boredom or envy.
  
  But a society with such machines can work just fine as long as we distribute the benefits fairly and do not allow a few people to seize everything.
  - manuel werner Dec 20, 2017
    
    Thanks for the reply. Unfortunately, history has no precedents when 'benefits' were ever fairly distributed and the Roman grain dole, evocative of the Universal Basic Income schemes we hear about today, did not end well. If you have a moment, I did write a short piece on this matter, at: http://www.theabelardchronicles.com/2017/10/automation-and-the-roman-grain-dole-a-cautionary-tale-for-universal-basic-income-schemes/
    A new reply to this comment has been posted.
  A new reply to this comment has been posted.
A new reply to this comment has been posted.
Phil Cartier Dec 18, 2017

I took part in an automation project a few years ago. After tripling the productivity of the conventional process, which required a total of about 40 employees the company decided to build a new plant.
The new plant used a new process and equipment I developed which was intentionally not totally automated. No one trusted that. When installed the new production was 20% higher using at most 7 people. The gross margin was in the 60% range. After a few years plant productivity programs run mainly by the people on the line cut the number down to seven. Some added packaging automation reduced the number to 3, plus 2 people for about 3 hrs a shift.
All of the production workers involved were getting above average wages with high school degrees or less.. The key to making it all work was that the company valued the workers and selected smart people who were willing to work and learn and get along with others. Those abilities, more than anything else made to project work.

Unfortunately the workers in the prior plant who lost out were protected by a union and many had an ethos of I'll work as little as I can get away with.
A new reply to this comment has been posted.
Phil Cartier Dec 18, 2017

I took part in an automation project a few years ago. After tripling the productivity of the conventional process, which required a total of about 40 employees the company decided to build a new plant.
The new plant used a new process and equipment I developed which was intentionally not totally automated. No one trusted that. When installed the new production was 20% higher using at most 7 people. The gross margin was in the 60% range. After a few years plant productivity programs run mainly by the people on the line cut the number down to seven. Some added packaging automation reduced the number to 3, plus 2 people for about 3 hrs a shift.
All of the production workers involved were getting above average wages with high school degrees or less.. The key to making it all work was that the company valued the workers and selected smart people who were willing to work and learn and get along with others. Those abilities, more than anything else made to project work.

Unfortunately the workers in the prior plant who lost out were protected by a union and many had an ethos of I'll work as little as I can get away with.
- Paul Friesen Dec 18, 2017
  
  A good illustration of why we need to understand what is happening and do something. We cannot buck this trend, but we need to understand that the way things worked in the past is coming to an end. We need to stop trying to punish everybody into working as hard as possible. It just is not necessary any more. We need to develop ways to let those whose work is still needed work and get rewarded, but at the same time recognize that there are already people whose work is just not needed because it is done by machines. Those people should still get their share of the output of the machines.
  A new reply to this comment has been posted.
A new reply to this comment has been posted.
Dr CKB NAIR Dec 17, 2017

The proposal for Tax relaxation for life long learning to be implemented in all nations for equal development
A new reply to this comment has been posted.
Petey Bee Dec 14, 2017

For one thing, technology makes offshoring even easier. That's past tens of course, but the trend continues. For another, simply having less expensive tools and automation, and infrastructure system designs requiring less maintenance labor and so on will reduce labor across the board.

So this piece is a bit tough to reckon with observable reality, even before we get to the future of AI. Been to a supermarket lately? Automated checkouts. Rented a movie or bought a movie ticket recently? No? Ah, ok. Bought a financial service or insurance product at the retail level?

Self driving cars, for all their foolish side effects (put more vehicles on the road, impeding transport), are readily feasible with a few simple changes (putting little RFID transmitters in the road).

Lets keep going. How much did you buy this year from amazon or ebay? How much from conventional retailers? How are we doing on interactions with sales-people, accounts-payable people, or telemarketers vs phone-bots and electronic equivalents? Back to other "regular" jobs- Watched how modern construction is done? It has more and more power tools and less and less labor. Same for building and maintaining roads and electric/pipeline infrastructure.

And the coders, techs, and engineers who make this all possible, as well as their immediate managers? Modern comm, shipping, and translation tech allows them to be most easily offshored as well.
A new reply to this comment has been posted.
B Wilds Dec 13, 2017

Creating "real jobs" in a mature market or economy is not an easy task and poses difficult challenges. Globalization has elevated the importance of creating jobs and a balanced economy that supports a strong middle class.

We must differentiate the difference between creating a valuable and worthwhile product that benefits society and breaking a window then praising the jobs replacing it yields. The article below delves into what constitutes a "real job" and its value.

http://brucewilds.blogspot.com/2016/11True-job-creation-constitutes-real-challenge.html
A new reply to this comment has been posted.
Enrique Fleischmann Dec 13, 2017

“THE GENIUS of the industrial system lies in its organized use of capital and technology. This is made possible, as we have duly seen, by extensively replacing the market with planning” (JK Galbraith, The New Industrial State).

It is rather surprising that the observed and expected technological progress is limited to a supply side/ micro economic perspective: Substitution effects (Capital and Labor) is just one example to the biased focus. How to ensure a stable demand ( and a fair distribution...) to absorb the expected flow of products and services should be incorporated into the debate.

The typical analysis of the AI revolution departs from a sort of micro economic model applied to a macro perspective. This version excludes time dimension and any associated uncertainty. Such omission may be useful for analytical purposes; however, time and (un) certainty mark substantial differences between Capital and Labor, differences which deserve further analysis.

Very shortly: a Capital intensive economy requires, Ceteris Paribus, a stable level of demand. The huge investments in Capital (comparing to Labor) in technology are financed by capital owners which expect periodical yields ( interests, dividends, etc). Reducing uncertainty and competition is fundamental to ensure timely yields and profits.

Ensuring and regulating demand in a globalized economy is a task beyond the power of market players. The main economic agent able to ensure a stable economy is the State... so i guess that the Demand side of the Capital Intensive economy will probably bring about more state intervention ( Basic Income as an example) from the side demand.
- Steve Hurst Dec 13, 2017
  
  Enrique. 'Ensuring and regulating demand in a globalized economy is a task beyond the power of market players'. Yeah okay, what make you think a globalised activity can be effectively regulated locally. Or to put it another way - the WTO's future looks to be fragmenting effectiveness and it is just one international body with problems in the face of globalisation outcomes. The main containment of globalisation is likely to be demand - which has dropped hence the flattening of international trade. The problem with AI and robotics is it will follow the 'me too' model of offshoring - the ... I have to do it because everybody else is doing it ... justification. Chances are that will also reduce demand. To paraphrase the BoE comments in the Gt Recession re the financial sector - activity is multinational but liability is national - that still is the case it would appear. Capital has castrated regulation
  A new reply to this comment has been posted.
A new reply to this comment has been posted.
Ariel Tejera Dec 13, 2017

Never had humankind had more information available about its nature, and it's circumstances. We are perspiring distilled knowledge. Yet... we have no means of deciding what we want humankind to be. We are drifting directionless.
A new reply to this comment has been posted.
Michael Public Dec 12, 2017

What this article fails to see is that work consists of two things: force and intelligence. When they mechanised farms and industry people took jobs which were intelligence based. Now that pattern based intelligence is about to be automated there is precious few places we can go to add value. And the things you mention as outside the realm of AI are already on sale for example compassionate conversation AI driven bots that are popular in China.
- Steve Hurst Dec 12, 2017
  
  Micheal, importantly when the mechanised farms (40+% of the population initially, halving in a matter of a couple of decades; then heading onwards to the current <2%) the displaced workers could find employment in a new sector, manufacturing consumer goods. The question is what new major sector is there around the corner and the answer is it is not at all obvious
  A new reply to this comment has been posted.
A new reply to this comment has been posted.
Michael Public Dec 12, 2017

The economists' notion of productivity is lost as sea because it does not take into account changing prices. The phone I am typing this on would have cost a trillion dollars to make in 1990 and I paid 300 dollars for it.
- Peter Schaeffer Dec 12, 2017
  
  MP, You don't understand how productivity is calculated. TFP and GDP do take into account changing prices. However, the Tech sector just isn't large enough to drive TFP and GDP all that much. Cell phones are cheaper today than in 1990 (and do much more). However, housing, health care, and education are not. Of course, the details depend on what country you are living in, where you live, what you do, etc. The bottom line is that stagnation (and overt decline in some cases) are the reality of our time.
  A new reply to this comment has been posted.
A new reply to this comment has been posted.
Paul Friesen Dec 12, 2017

Two problems.
1) We don't have real artificial intelligence yet, but it is sure to come. Machines of the future will be able to do anything a human can do, including building more machines like themselves. Do you really think that will be just business as usual?
2) Even if you think that "productivity" increases will just allow the same number of workers to produce ever more stuff with the same labour, there is a problem. The environmental and energy costs of all the stuff we produce are already huge. Much of what we produce is already really surplus if you think about it. People own three cars because their neighbour owns three, not because they have any use for them or because it actually gives them more pleasure. How far do we take this? Do we ever say we have enough?

There is a solution. We could pay a basic income, paid for by sales taxes on all that stuff. It would reduce inequality and allow people more leisure, while reducing our environmental footprint. We take the payoff for all that "productivity" in leisure time, not more and more stuff.
A new reply to this comment has been posted.
Steve Hurst Dec 12, 2017

Here in the UK retail bank outlets are shutting down left, right, and centre. Within the structure left staff are being trimmed left right and centre. Just to give one example. They are being replaced a combination of tech - the internet, smart phones and tablets and AI. Those jobs are not being replace within banking. The entire purpose of AI is to reduce labour content and the assertion lost jobs will be replaced is just that, an assertion.

In another sector - Healthcare; there was pioneering work done by a doctor in the US decades ago on software based diagnosis of ailments. I cannot remember the guys name offhand but he considered his software was better than he was due to the entirely logic based process and consistency of the process. It should be noted that healthcare is subject to very significant AI investment at present.

Many businesses are poor at investment in personnel and certainly poor at investing in staff they have sacked. The German model referred to is basically the continuation of the medieval guild structure and jobs are legally protected. You cannot for example trade as a piano tuner unless you have undertaken and completed years of apprenticeships. The idea this could be imposed, which is the word, imposed, on businesses in the US is fanciful not least because it inevitably puts the cost up to the consumer who will baulk. We have the bizarre situation that cheap migrant labour to undercut costs is forever championed as being essential for the economy to function and at the same time the claim skilled jobs will somehow be protected when there is no evidence that will happen. As for strong unions, that is the entire focus for destruction in most developed economies. It has happened in the UK and implicit in the election of Macron is an attack on public spending and therefore the powerful French public sector unions. The so-called 'Gig' economy is based on employers wants not employees needs. Itis ironic that Uber drivers in what is very much a Gig economy activity actually support Uber in any attempt to limit Uber activity by regulatory authorities when Uber have quite clearly and repeatedly stated their near horizon objective is the sacking of all drivers and to use robo cars.

I'm sorry BE but I just dont buy your argument
- Phil Cartier Dec 18, 2017
  
  As an alternate, I just read an article about a very successful developer of shopping malls. His secret sauce is that shopping is a social experience with the sales stat that many people enjoy. His malls cater to that making shopping an emotional experience that draws people in.
  The article also posted some figures showing that overall retail shopping is well over 90% in store. Online is growing, but even Amazon is going retail. Online is well under 10%. Perhaps the next Amazon will figure out how to make online buying a social, emotional, satisfying experience.
  A new reply to this comment has been posted.
A new reply to this comment has been posted.
Keith Roberts Dec 12, 2017

I have great respect for Prof. Eichengreen, and always read his PS articles. But in this case, Homer nods. Of course there is plenty of need for human workers, a need for human contact, companionship, and help that robots cannot fill. But we have not devised a way to pay a living wage for those kinds of job. And no, factories of robots will not replace all their workers. The robots need oiling, after all. But to say that the nature of work will shift does not take the reality into account. To use the lawyer example, if a law firm can use AI for most of its research, just how many associates will it need? How many "secretaries" are there any more? Prof. Eichengreen is trying to reassure us, but in doing so he is not addressing a real looming problem.

Nor do I find his historical statistics reassuring. The normal pattern of innovation growth is a one-tailed parabola, with a long period of very little happening, and then a rapid acceleration. We seem to be at the very beginning of that period of rapid acceleration, so the preceding stagnation tells us nothing useful.
A new reply to this comment has been posted.
Peter Schaeffer Dec 12, 2017

Companies own machines but don't "own" workers. Investing in workers doesn't pay off (in the USA) because of personal and occupational mobility. Stated simply, the company spends the money, the worker owns the property (the newly acquired skills in this case). The solution (in the USA) is for the government to provide the funding. In real life, this used to work better than it does not. At one time, public schools taught "shop" where high school kids were given useful skills. These days "shop" isn't PC because "everyone" should go to college.

Reality hasn't changed. However, the elite fantasy that everyone can do as well as their kids (attend an elite university) has become a standard (fantasy) worldview.
A new reply to this comment has been posted.

https://prosyn.org/94V4c0R

Two Myths About Automation

PS Events: What Economics is Missing

New Comment

Contact us

Handpicked to read next

Rage Against the Machine?

Log in/Register

Two Myths About Automation

PS Events: What Economics is Missing

Handpicked to read next

Edit Newsletter Preferences

Set up Notification