Automation and the Future of Work

My mother sent me this TedX video of David Autor, an MIT Professor of Economics.  It's great.  I highly recommend it. Here is his paper on the subject offering even more detail.  

The impact of technology is all around us and just seems to accelerate leaving entire generations in lower-paid, less skilled jobs than they had only 30 years ago.  This chart clearly shows waves of losses and gains in US employment by sector between 1940 and 2010.

So technology has utterly transformed how we work in the past century.  Although you can see the drama in these numbers and imagine the disruption to our society, is this really a problem?  Or is this a good thing?  In fact, this is an older, global process, arguably going back to the industrial revolution of the 19th Century and financial, commercial and trade revolutions reaching back to the end of the Middle Ages.  But what's the matter with this?   Aren't the rewards clear enough to see? 

In fact, this is a paradox:  an APPARENT (but not real) contradiction.  Development overall shown in the chart above has not been bad for most of us.  In fact, there is no real controversy here:  although there have been real costs to some, most of us would recognize that we all have a higher standard of living and would agree that the benefits to all of us far outweigh the costs.  

How does this actually work?  Why hasn't it been a zero sum game where the machines take over, at least not until now?  

Autor, a professional economist, does an excellent job explaining how this is possible.  He begins by recognizing the virtuous cycle of technological development and labor in a market economy.  First, the more we invest in technology the more productive we are as an economy.  He's looking at labor in the aggregate.  Greater wealth stimulates demand. Displaced labor finds even more valuable employment opportunities, acquires new skills, earns more and then demands even more goods and services. Growth spawns more automation. It's all good.

But Autor also concedes there is a challenge:  unless we apply the wealth generated by automation to our institutions, to our society, and to ourselves, the virtuous cycle will end.  Although he does not even mention inequality in his TED talk, his written work does acknowledge that it might be severe enough to limit broad-based consumer demand, breaking the cycle. Again, there is no controversy here.  Our academic, political and business leaders agree that these risks are real, even if we disagree on how likely they are or how severe they might become.

I would break his argument down into four basic parts:

1. Automation makes human labor MORE valuable:  the 'O-ring' effect
2. Productivity creates wealth that spawns new demand:  we can't get enough
3. Human ingenuity has always saved us in the past:  this time is NOT different
4. He's telling us something new, something we have not heard elsewhere

Let's critically examine each of these in turn.

First, overall, automation HELPS human workers be more productive, it does not replace them.  The 'O-ring' is a metaphor for the value of even the least expensive critical part in a complex system:  the more reliable the rest of the system becomes, the more value there is in every critical component.  Of course he's right:  SOME of the remaining jobs are indeed even more critical to the system as a whole and become more valuable.  For example, there are more bank tellers now than there were before the advent of ATM's and, overall, they are paid more and offer more complicated, more valuable kinds of customer service than their robotic 'assistants'.

On the other hand, there is plenty of evidence to the contrary that suggests that MOST of the remaining jobs are, in fact, less critical to the overall systems performance and productivity and are therefore generating less value.  Sure, there are some good jobs at Amazon, but they aren't nearly as numerous as the retail jobs they have displaced.  Even though Walmart has grown, retail sales in that context no longer require or pay for the skills it once did.  So these effects reducing the value of human labor are also real, offsetting the gains in other jobs. I would love to see these kind of labor statistics across the entire planet, the only way to study the effects of automation in a global economy where immigration and globalization negatively impact earnings in addition to automation. But I've not seen those studies.  Even without the hard data, I think we can agree that both automation and globalization play a role and that reality is more complex than Autor would have us believe with his 'O-ring' argument.

And so far, there is no news here.  These are old, old stories that are not as controversial as Autor suggests.

Let's consider the argument that productivity creates wealth that raises our standard of living and in turn spawns demand. Does it?  Well, of course there are no limits to what humans want.  However, if we need money from our labor to fulfill those desires and we're limited by how much our labor is worth, there ARE LIMITS to how much demand can be expressed in consumer markets.

As Autor says, "It depends on what we DO with the wealth."  If most of that is captured by the the top 1% and saved, well, it simply does not stimulate demand and growth.  Even if we invest in institutions like education as we have in the past, what jobs would we be training our displaced workers for precisely?  It's not clear.  There is plenty of evidence that that wealth is indeed being invested.  It must go somewhere.  But without demand,  all that capital is being applied to stock buybacks and acquisitions, not growth and not new jobs.   It seems to me that we have a surplus of capital and capacity which is consistent with disappointing growth in high quality jobs.

Look at the reduction in what we would consider "middle income" jobs:

He looks at this data and sees the glass half full.  I look at it and worry, not just about the jobs themselves but about how this affects the distribution of income and the engine that drives demand and growth.

What we have instead of broad-based participation in the automation economy is what Autor calls "income polarization," another name for inequality:

Again, no real news.  In his TED talk, Autor makes an oversimplified argument which is true... but incomplete.  Why not concede that there are also mitigating factors that diminish the effect of productivity gains, specifically increased inequality? 

Finally, what about human ingenuity?  Perhaps I am unimaginative and perhaps we WILL be saved.  I just don't think so because this time IS different.  Of course, every time this has happened in the past it HAS ALSO been different.  But we need to pay attention to WHY SPECIFICALLY this time is different and look for differences that might affect the balance of displaced jobs and income inequality I describe above.

Consider:

1. Until now, the vast reservoir of new work that has absorbed displaced labor has been COGNITIVE work.
2. Until now, the growth in COGNITIVE labor has increased the wealth of the middle classes which has also increased DEMAND.
3. This time the machines are COGNITIVE machines.
4. This time the machines are NETWORKED, creating essentially a singular, global machine.
5. Digital communications are essentially free, connecting everyone on the planet to the global machine.  This time there are fewer essential human ‘O’ rings.  This time there are even fewer owners reaping the rewards of automation.
6. This time, more displaced workers will make LESS income, not more, REDUCING DEMAND. 

Ultimately, it’s true that we “can’t get enough.”  Unfortunately, that unlimited demand is only expressed in markets when those people have money.  And increasingly, it’s only the people who are making the machines and ultimately OWN shares in the machines who are collecting all the money.  Everyone else is UNDEREMPLOYED but working harder than ever. That means they are earning and consuming less.  This leads to a crisis in DEMAND.  Which leads to slow or negative growth.  See Robert Reich's "Inequality for All" if you haven't already;  it's fabulous.

But as Autor says, this is not PREDETERMINED.  We can choose to redistribute wealth in new ways to prevent the worst inequality, to share more widely the benefits of automation, to enable more people to work less.  By sharing more, by allowing more people to participate in the automation economy, we might even stimulate demand while we elevate our standard of living.   

But will we?