The Curious Task: The Automation Problem

1
883

[Fair warning, this post is a bit long. This is a transcription from a Podcast Episode. If the post is a bit long for your tastes, please consider listening to the podcast episode instead.]

Does automation kill jobs? 

Since 1900, Manufacturing and Agriculture jobs have diminished a great deal. Using a measurement of total US employment, we can observe there has been damage to employment in these industries. Agriculture went from almost 40% in 1900, to less than 2% by 2010. Manufacturing from 25% in 1900, to just under 10% in 2010. 

While automation is not responsible for the total of the losses, it has acted as an accelerant in many positive ways. Since 1990 manufacturing employment fell by 30%, and output grew by 71.8%. In 2016 the United States produced 72% more goods than it did in 1990, with only 70% of the workers. Over the same period of time, manufacturing productivity in the United States grew by 140.1%.

It’s easy to hear these numbers and have your mind made up. 

Source: NCCI study on the impact of automation on employment. 2017. 

Manufacturing: The data isn’t what it seems, automation is just the fall guy. 

The numbers look staggering and too obvious to ignore. Manufacturing unemployment is up, productivity is up, it’s got to be the robots, right? Wrong. Manufacturing job growth hit its peak in the 70’s, since then job growth in this area has been in decline. The only turn-around that’s been seen of yet on this trend, is following the great recession. Employment in the manufacturing industry actually began to increase again after 2010. 

The recovery has been sluggish, to be sure, and it’s far away from the golden days of manufacturing employment, but it’s a recovery nonetheless. And the fact that this recovery exists, goes against all conventional hypotheses supporting the automation argument. 

The China problem

And something important is recognized in the data between the 1970’s and 2010, and that’s a substantial amount of manufacturing jobs brought to the guillotine in 2000, to the tune of about 5 million jobs lost. This is an all too common talking point for those warning of the dangers of automation and its ability to kill jobs. But these jobs were not lost to robots. In 2001 China joined the World Trade Organization, and Chinese exports to the USA saw a sharp increase. Trade researchers have even come up with a name for this, and they refer to it as “China shock.” I’ve never been one to put much importance on trade deficits, but in some instances it’s the only data we have. In the decade following China’s entrance into the WTO, US trade deficit as a share of GDP averaged -4.3% of GDP. Dating back to reports starting in 1930, that’s the highest negative spanning a decade on record. 

I don’t personally believe tariffs are the answer here, but I think that in light of this evidence the opinions of Trump are a bit more accurate than those of Yang and co. The truth is, China is able to produce what it does due to unsustainable practices and policies. As a result, the Chinese government is now pumping the brakes on a large amount of its steel industry as it exercises all of its tools available to centrally adjust the direction of steel production and coal mining. 

Larger issues with other international trade partners lay in frail standing as the urgency to fix this industry problem increases for China. For one, current levels of production go against China’s proclaimed focus on green initiatives. China also has a significant amount of “zombie” operators, resulting in production to the tune of an estimated 300 million tons of excess iron and steel. 

Clearly this is a difficult position for any international trade participant, but it’s made entirely more difficult when one tries to match what has been achieved by China in this regard. Sometimes the best offense, is a good defense. In the face of these challenging market factors, further regulation and and oversight only seeks to place an already disadvantaged industry at an even larger disadvantage. Preventative measures can be taken to allow these industries to remain more agile and competitive while waiting out China’s inevitable (and already in process) industry reform. 

It’s not just China, either. 

In a study released January 2018 by economist Susan Houseman of the Upjohn Institute, it’s documented that when you remove one specific portion of the manufacturing sector, everything changes completely. That specific portion is computers and electronic products, which add up to 13% of manufacturing output. If you take this area away from the industry, it’s quite clear that manufacturing output has actually been lagging behind the rest of the economy.

Computers and electronics account for such a large portion of output in the manufacturing industry because higher output in this area results in lower prices and more productivity. However, the lower prices are due to better and more efficient and effective tech components, and have nothing to do with process automation. Components being more effective and less expensive doesn’t mean there are less technicians available to install them.  Automation in computer production does raise GDP growth in the factory sector, but it doesn’t [or hasn’t, in the computer and electronics category] eliminate jobs. 

There may very well be areas where it’s unequivocally proven that automation kills jobs, and I think there’s a fair argument to be made to that point in the farming industry, but it’s not the culprit that it’s made out to be in the manufacturing industry, and this talking point should be put to bed. Having said that, automation certainly can, and does displace jobs in other industries, but this should be analyzed properly, and carefully, on a case by case basis. 

So, some jobs are displaced. What else does automation do to the job market?

There is another side to the nature of automation, and that’s creating new jobs. Since 1990, the total non-farm workforce has grown 33 percent. This more than accounts for the previously stated losses to the manufacturing jobs lost to automation. 

If automation killed jobs and that was the end of it, ATM’s would have surely given us a sharp decline in bank branches, right? Well, that’s not what happened at all. In fact, the number of available jobs at banks has increased, substantially. Employment didn’t fall in this industry because the ATM made it possible for banks to operate branches at lower costs. This resulted in the opening of more branches, branches that all required a headcount and amounted to new jobs.

A recent study by the World Economic Forum has estimated that by 2022 automation (machine learning and algorhythms) will create 133 million new opportunities for employment, but these opportunities will come at the cost of 75 million displaced jobs. Another study by PwC determined that automation will bring significant net economic benefits, to the tune of $15 trillion contributed to global GDP. 

A report by think tank Centre for Cities estimates that 1 in 5, or 3.6 million British jobs will likely be done away with by 2030 thanks to automation. Also, according to the report, despite the likely probability of job losses, there will be more overall jobs available by 2030. However, these will be jobs that require a higher level of cognitive and interpersonal skills. 

Also included in these studies is information about what kind of tasks or what kind of work will be automated… 

What kind of work is automation going to replace, and what kind of jobs will be affected most?

In a study conducted by the McKinsey Global Institute, the categories of activities in jobs were ranked by time spent across all occupations studied, and held against potential for automation to affect those jobs. 

Here’s a quote from NCCI Insights, about the McKinsey Global Institute Study:

McKinsey aggregated its 2,000 work activities into seven broad categories. This study shows the average percentage of time spent on each activity category across all occupations, as well as its potential for automation. For example, a 69% automation potential for processing data means that over two-thirds of the time currently spent on this activity by human workers across all occupations might be saved by automation with existing technology.

McKinsey estimates that only 7% of time is spent managing and developing people, and that this is the work activity with the lowest automation potential of 9%. The next three work activities—applying expertise, interfacing, and performing unpredictable physical activities—make up 42% of time, with automation potentials ranging from 18% to 26%. The three remaining categories—collecting data, processing data, and performing predictable physical activities—comprise 51% of all work activities and have high automation potentials ranging from 64% to 81%.

What this study found was there are certain types of work that are much more likely to be automated than others. The findings indicate that categories of work that are most likely to be replaced by machines is work that folks by-and-large don’t particularly enjoy in the first place. The highest-ranking category of work in terms of potential to be offset by automation are listed as; Performing physical activities and operating machinery in predictable environments (81%), processing data (69%), and collecting data (64%). After that there is a sharp decline in probability given to automation replacing jobs, the next closest category being “Performing physical activities and operating machinery in unpredictable environments” (26%). 

Work in this category breakdown that is seen to have a lower percentage chance to be influenced by automation include; Managing and developing people (9%), applying expertise to decision making, planning, and creative tasks (18%), and interfacing with stakeholders (20%). So, what we’re seeing is less of a need for low skilled, repetitive information collection/compilation, and manufacturing assembly line jobs. 

When we mirror this study with a similar Oxford study, results speak to a similar sentiment, one that strengthens the case made with the findings in the McKinsey study. We see what jobs have large amounts of the work described in the McKinsey study, and can then estimate which jobs would be most likely replaced by automation. The findings stated that jobs in categories such as; service, sales, office and administrative support, production, and transportation and material moving are high risk for automation replacement. Job categories such as; management, business, finance, computers, engineering, science, education, legal, community service, arts, media, and healthcare are much lower risks for automation replacement.

The jobs being replaced are jobs no one even likes anyway. 

What’s interesting is, according to research provided by the site number8.com (sourced from Gallup World Polls), if we measure job type / happiness, we can see the kind of jobs or tasks that are being replaced by automation, are actually the jobs that people have reported make them the least happy. As it turns out, the jobs that people indicate are the least satisfying, and lead the least amount to personal happiness and fulfillment, and the jobs that are the most likely to be replaced by automation. Farming, transport, manufacturing being among the lowest listed.  The jobs that tend to make those working them the unhappiest, are the same types of jobs that are most likely to be replaced by automation. 

Another argument, (I hate to keep going back to Yang but he’s mister scary automation at the moment) is that as a result of the effect automation has on jobs this leaves the workforce in a bad place in terms of mental health. The tough part to swallow with that argument is that all of the evidence we have seems to indicate the opposite. Automation generally leads to higher wages and increased job enjoyment for those still employed, and the new jobs created are the types of jobs that score far higher on any reports thus far available examining happiness in terms of profession. 

No one likes big changes, and that’s something most people are aware of and this shouldn’t be an argument. But personal responsibility has to come into play at some point. Making a change to a job you enjoy and get paid a lot for, from a job that you hate and get paid very little for is probably one of the least damaging changes an individual will have to experience. Sometimes our fear is greater than the threat, and I would wager that this is one of those times. 

Having said that, I’d hate to be lazy and assume this situation would fix itself. Political ideology aside, there are approaches that can help bridge the gap with job displacement. 

What can we do to counteract displaced jobs? 

What can be learned from going over all of the data we just have above, is that with automation, there is a shift in the way the market works. If it’s a shift that can be properly adapted to or accounted for, it can be largely advantageous to all involved. If it’s a shift with little possibility for adaptation, those who lose jobs could suffer as they struggle to fit in with a marketplace that has deemed their prior career up to this point, no longer needed.

What is the change that’s going to stick? The first place most people go is job retraining programs, and this approach is about as right as it is wrong. Clearly, learning new skills and being trained in new ways to operate inside a changing market is going to be an extremely valuable solution here. The problem is, to date most federal job retraining programs have proven largely ineffective. Often times the kind of jobs that these programs train for, and jobs in industries that are in decline, or jobs in industries with little opportunity. So what could be done to improve the state of job retraining programs?

1. Identify real shortages in proximity to those being retrained.
Not every job retraining effort fails in its attempts. Capital IDEA is an example of a training program that actually has a lot of success to boast, including; over 1,600 students placed in new jobs, and 3x increases in earnings to the program’s students. The key to success for Capital IDEA is the way it addresses market needs. They do this by working with educators and employers to identify where labor shortages exist, and then trains students as according to those requirements. If more retraining programs focused specifically on labor supply and demand gaps in local markets, one would assume they would find more success as a result, just as Capital IDEA has. 

2. Barrier to entry from institutions 

When someone loses a job, the last thing they have is the time to jump back into the education system with both feet. In this situation an individual is going to need a job in the immediate, not two years down the line. More “boot camp” type approaches toward building skills have come about lately, and I suspect these will continue to be as popular, as they are effective. Alternatively, companies are finding more and more that the most valuable employees can come from within, or without a 4-year degree. Companies as large as IBM have already began initiatives to help solve this problem, with the introduction of their “new-collar” program. The goal of this program is to identify shortages in the labor force as according to needs in the tech industry, and open doors for entry-level positions that don’t require expensive certificates and degrees that take years to acquire. David Leaser of IBM: “Right now, as many as one third of our employees at IBM have less than a four-year college degree.” Of course, things like education will always come into play, but companies are already identifying ways to address the barrier to entry problem, and it’s working. The more successful these efforts are, the more of them we can count on seeing.

3. Occupational Licensing
A study conducted by the Brooking Institution’s Hamilton Project estimates that licensing requirements that discourage people from pursuing careers have resulted in approximately 2.85 million fewer jobs nationwide. These days, around 30% of American workers are required to have a license to perform their jobs, compared to the 5% that existed in the 1950’s. The study identified around 800 occupations within the United States that are licensed by at least one state, many of them are required by all. 

The Truckpocalypse

Another big concern right now surrounding discussions involving automation, is the trucking industry. While the trucking industry is being pursued by those leading innovations in automation, the future isn’t a bleak as is often made to be. The trucking industry isn’t going to just save money by slashing workers salaries, that’s not how this works and it’s not close. Automated trucks are actually much safer, a safer truck means lower insurance premiums and less paid to insurance or paid as the result of accidents. Automated trucks are also able to take advantage of advanced driving methods that save on gas mileage, this method is called platooning. Two semi trucks platoon at a constant 64mph, and a 36-foot following distance. This configuration leads to an average fuel consumption saving of 4.5% for the lead truck, and 10% for the following truck.

And what about the drivers jobs?

There are 5 levels of automation for trucks. Level 0 is no automation at all. Level 1 is driver assistance, at level 2 the vehicle can control acceleration, breaking, and steering in prescribed situations. Level 3 is as far as we’ve got so far with automation and vehicles, no level 3 production cars exist on the highway as of now. Here the car does most of the driving, but a driver is still needed to pay attention to road conditions. Level 4 trucks could drive themselves, but would still need pilots to be in the vehicles, however, these pilots could switch out at freeway junctions, letting truck drivers stay closer to home, making long weeks away unnecessary. At level 5 the driver is now just a passenger, monitoring performance during the journey. This is where we need to get before trucks can go from depot to depot without human interaction. The industry currently predicts that level 5 is still decades away.

The truth of the matter is that there is still a real need for class a qualified drivers, in fact the industry is currently at a shortage. It’s a currently unmet market need, and has been. If it remains as it has been, level 3 or 4 will be welcomed by the industry as it will help to bridge the gap.

The Tertiary Affects of Automation

Automation displacing jobs also has an effect of creating a market, in a sense. The more jobs are displaced for automation, the more things become automated (obviously) and are more assembly line than they are personal creation. With this surge in assembly line goods, there also comes a market demand in products that still have a more personal feel. We can see this clearly evidenced in the rise of platforms like Etsy, Restaurants boasting organic and farm fresh ingredients far removed from the automated mechanized farming techniques automation is responsible for, or the sharp boost in small wineries and breweries, where a personal approach is a big selling point.

Automation is absolutely possible in these areas, but companies have started to use the lack of automation as a value proposition, and it works. Automation makes products cheaper, and more abundant, and as a result people naturally find ways to shift spending to goods and services that still have connections to human providers.

What you can count on seeing more of are workplace policies like those put in place by Starbucks, who stated baristas were no longer to make multiple drinks at a time. This resulted in longer waits, slower lines. But it also added to something else, and that’s the experience the customer has. The goal was to assert to the customer that no matter what, their barista devoted undivided personal attention to preparing their drink. 

Economists like Andrew Yang, and many others will be quick to address the issue of automation by simply throwing money at it in the form of a UBI. The idea with these proposals is to paint a grim picture of the future of the US job market, and posit that the only solution available is to steal from Peter to pay for Paul in the face of this automated juggernaut. Andrew Yang himself views opposing free-market opinions as “lazy” or “shallow” but I would argue that his own approaches are far more of what he Is accusing others of. 

Properly addressing labor gaps, deregulating the occupational licensing landscape, adjusting policy to allow industries to remain flexible and competitive in the face of international mercantilist forces, and finding incentives to lessen the burden of institutional barriers to entry are all far more effective than a UBI would be, something that we can address at a later time, on another episode. Automation is cleaning up the industry, taking away the monotonous and labor-intensive tasks no one was exceptionally excited about in the first place, accelerating productivity, and creating opportunities where they simply did not exist prior.

The market requires more bravery and less fear, as we adopt more flexible methods of responding to market signals, we will only see benefits. And when doors close in the market, new and innovative doors open, what automation replaces and makes more efficient can become a market demand. In the face of an increasing volume of impersonal economic activity, a higher value and opportunity will be given to those who can bring a more personal approach.

The workforce needs to respond by remaining flexible in the face of changing markets. That is the *only* answer to the automation problem. The sooner that everyone is able to understand this, the sooner solutions can be held up that are far less lazy than throwing money at people and manipulating an already broken system of incentive structures within the US economy. 

The following two tabs change content below.

Vinny Marshall

Vinny is a digital marketing and design professional, and an active voice in the Liberty Movement. A frequent contributor to Being Libertarian, owner and operator of Anarchothreads, co-Founder and frequent contributor of Think Liberty, and host of the Think Liberty Podcast.