Our Robot Economy

Donald Trump blames China for job losses in U.S. manufacturing, but he should blame robots. Ball State University reported that trade and offshoring was responsible for 13% of manufacturing job losses. The remaining 87% were lost to automation. McKinsey has suggested that 45% of the tasks people are paid to do today could be replaced by existing technology — this doesn’t account automation enabled by fully self-driving trucks, for example.

Trump’s initial (later withdrawn) nominee for Labor Secretary, Andrew Puzder, understands the appeal of automation well enough. He explained the virtues of robots versus human employees to Business Insider: “They’re always polite, they always upsell, they never take a vacation, they never show up late, there’s never a slip-and-fall, or an age, sex or race discrimination case.” Puzder would know, having run some of America’s largest food chains as CEO of CKE Restaurants.

We are terrible at predicting the future, but it’s almost certain that increased automation will result in less work for people. It’s less certain — though still a reasonable bet—that in the near future robots will be able to do most of our work for us.

Some people are skeptical of robots and their capacity to replace human labour. Matt Yglesias makes this case in Vox. The argument is basically that automation has been happening in various forms for a very long time. Now, like in the past, new jobs will pop up to replace the old ones now done by machines. How this replacement happens is never quite explained, but that’s what has always happened before and so it’s what must happen in the future.

I think this skepticism radically overstates our own irreplaceability and understates the gap between the machines of the Industrial Revolution and the artificially intelligent machines we are now creating. Robots are already better than people at doing most mechanical jobs, and with some basic AI built in it turns out they can also do higher-level work like design, legal analysis, and medical diagnosis. Watson, for example, has already made correct diagnoses in cases that doctors haven’t been able to crack.

Even in what are widely viewed as “growth industries” like tech, opportunities are not what they once were. A small number of highly-skilled programmers — the people making the job-eating software — are in-demand and highly-paid, but lower-level programming work is drying up. Software like Squarespace is already replacing work that used to be done by junior web designers and developers. Governments recommend skills re-training, and many people dutifully shell out tens of thousands of dollars for “coding bootcamps” only to find that these skills no longer guarantee work. The re-training panacea that many governments promote is a pointless exercise in delaying the inevitable realization that there is simply not going to be enough work to go around.

Contradictions in the Robot Economy

One of the effects of automation is that the share of national income captured by labour will continue to fall, probably more sharply than it already has. Here’s The Economist back in 2013 explaining the drop of labour’s share of income in developing countries:

Workers in many developing countries, from China to Mexico, have also struggled to seize the benefits of growth over the past two decades. The likeliest culprit is technology, which, the OECD estimates, accounts for roughly 80% of the drop in the labour share among its members. Foxconn, for example, is looking for something different in its new employees: circuitry. The firm says it will add 1m robots to its factories next year.

As labour’s share of income falls, capital’s share of income will rise. The majority of capital income flows overwhelmingly to the top 1%, as the chart below illustrates.

This will exacerbate the wealth gap between the owners of capital — the 1% — and everyone else. But beyond that, automation is also likely to drive down wages in general as more people compete for fewer jobs. The few highly-specialized workers who remain employed will be even more well-compensated as automation allows them to generate more value for their employers. These 1%-ers who have high labour incomes will, along with their capital-rich employers, continue to rise faster and further than everyone else.

This poses (at least) two problems for democratic, capitalist societies.

The first is that automation undermines the myth of the striving entrepreneur which is the last line of defence for capitalists justifying vast inequalities. The truth is that most of the very rich would still be very rich if all they did was lounge by the pool and collect rents, but the myth is apparently believable enough to legitimize capitalism to most people. This is essential if capitalism is to co-exist with democracy — any economic system that isn’t widely seen as legitimate in a democratic society will eventually be discarded.

Robots will heighten this tension between capitalism and democracy. Matt Bruenig explains why:

Robotization would dramatically shrink the only kind of income (labor income) that fits squarely within the prevailing economic ideology. In so doing, it would also increase the need for welfare incomes that are going to be hard to describe as earned and also hard to denigrate as essentially bad. At the same time as it is dramatically shrinking the workforce and increasing the welfare rolls, it would also dramatically increase the share of income going to capital, which would unavoidably put capital income at the new center of the economic discourse. In these circumstances, the trivial fact that robot owners do not do anything for their riches would be very difficult to ignore and even more difficult to justify or explain away.

The second contradiction between capitalism and democracy in the robot economy is that political power will accrue to a tiny number of rich people. In countries with lax campaign finance rules, like the United States, this is achieved through formal channels of direct contributions and election spending. The rich have already captured most governments at the state and local level in America, electing thousands of hand-picked stooges who reliably promote their interests. The Presidency is more competitive, but the donor class exerts a great deal of influence over the nominees of both parties.

In countries with stricter campaign finance laws, the rich exercise influence less directly but with comparable force through informal mechanisms. This piece in Jacobin outlines the ways in which the wealthy are able to constrain and influence policy decisions without direct intervention — to summarize:

Capitalists routinely exert leverage over governments by withholding the resources — jobs, credit, goods, and services — upon which society depends. The “capital strike” might take the form of layoffs, offshoring jobs and money, denying loans, or just a credible threat to do those things, along with a promise to relent once government delivers the desired policy changes.
As the robot owners accumulate a greater share of national income, their ability to direct policy will become even greater — and no doubt they will shape it to suit their interests. We can expect a vicious cycle in which the rich are able to capture government in order to further enrich themselves and in so doing solidify their grip on policy.

Capitalism and democracy have always had an uneasy co-existence filled with tensions that governments have tried to smooth with the welfare state. To some extent this has been successful. But automation will heighten these contradictions dramatically, and push the system to its breaking point. The future is going to look radically different than the present — the question is what sort of future it will be.

Three Futures

Until now I have argued that democracy, capitalism, and automation are an incompatible trifecta. Only two can co-exist. That leaves us with 3 possible configurations.

The first is capitalism and democracy. This is the status quo, or perhaps would have been considered the status quo prior to the 1990s. For reasons I have already argued for, capitalist democracy demands the sort of economic conditions that preclude domination of politics by the rich. In practical terms this means divorcing money from politics in both formal (strict limiting of campaign donations) and informal ways (reducing inequality from today’s levels, breaking up overly-powerful monopolies, and pursuing other policies that limit capital’s influence).

Preserving the conditions that make capitalist democracy possible would entail limiting automation, and possibly rolling it back somewhat even from today’s levels. The irony here is that capitalist democracy is what many conservatives and neoliberals want, the same people who often accuse others of Luddism. And yet their own politics demands technological backwardness unless they are willing to give up on democracy (something they usually won’t fess up to).

In any case, this option is impractical for all sorts of reasons. I doubt we could effectively limit automation, but even if we could there are strong reasons not to do so. Automation within a redistributive economic framework will increase people’s leisure time and quality of life, a change we should welcome. If your priority is democracy, then why toss out automation to preserve capitalism when the former delivers the goods in a way the latter cannot? And if what you want above all else is capitalism, then limiting technological progress should be anathema. In either case, I do not think this is a tenable option. We should make other arrangements.

A second possible future includes capitalism and automation. This sort of society would be dominated by oligarchs who have accumulated vast wealth. It could maintain democratic theatre, or perhaps look more like old feudal states with lords and peasants. Societies like this may also choose to exclude all except for the economic elites, something like the offshore sea-steading states envisioned by Peter Thiel and other libertarians. How this type of society functioned in practice would be a factor of the ideological apparatus constructed to justify this configuration, and it’s not yet clear to me what that would look like.

This sort of society would likely have greater wealth inequality than our own, and fewer (if any) redistributive policies to mitigate that inequality since the people who demand redistribution would lack political power. Most people would live in poverty, either unable to find work or competing with everyone else for the nominal wages of work yet to be automated.

If you find this vision of the future disturbing, you should consider that this is the default. Without structural change, automation will funnel more wealth to a tiny number of people who will come to dominate politics. Shades of this future are already visible in the United States where money is legally treated as speech, and has been used to aggressively (and successfully) promote policies favourable to the 1%. As this process accelerates, the ability of people to resist the consolidation of power through normal democratic channels diminishes. Third-party analysts already consider states like North Carolina, which legally disenfranchise vast numbers of voters, only partially-democratic. Unless people push hard against this current, expect it to continue to pull us further from actual democracy as automation accelerates.

The third configuration is democracy and automation. This is my preferred outcome, as it preserves actual democracy without giving up the benefits of automation. Limiting the political power of robot owners is the key to achieving this future. I see two ways this can be done, though there may be more.

One way is to maintain existing capitalist styles of ownership but redistribute the wealth created by robots to everyone rather than allowing it to flow to just a few. This would involve significantly higher taxes on capital gains and income of the rich. I’m not sure what the exact level ought to be (and I would prefer a basically equal redistribution), but something that gets us to post-World War Two levels of inequality would probably be sufficient. This option would also require controls to prevent capital flight and tax evasion. Whether these could be implemented successfully is an open question that needs more exploration.

A second possibility that skirts the problem of capital flight is broadening public ownership of capital. Rather than a few dozen people owning the robots, the public owns them through a vehicle like a Sovereign Wealth Fund. Alaska’s Permanent Fund is an example of this model in practice, though on a much smaller scale. In the 80s the state established the fund and began to finance it with oil revenue. The Permanent Fund is now worth $57 billion and pays out a dividend to every Alaska resident each year (in 2015 this was over $2,000).

Sweden has experimented with different models of bringing private capital under public ownership. Wage earner funds established in the 80s were owned by employees and financed by dedicated corporate taxes. They eventually grew to encompass 7% of Sweden’s economy. In the 70s, the Meidner Plan called for the creation of similar (but more expansive) Workers Funds. Under this proposal, companies greater than a certain size would have had to distribute shares to its employees equal to 20% of the year’s profits. Over time this would have led to worker ownership of large chunks of every significant corporation in Sweden.

You could capitalize the fund in other ways: new taxes on capital, counter-cyclical asset purchases, scrip taxes on corporations that pool a small amount of corporate shares in the fund each year, and higher inheritance taxes on large estates are all possibilities.

The point is that there are a number of viable and realistic ways to bring capital under public ownership, all of which would reduce the power of economic elites and protect democracy from their influence.


I have argued here that robotization leads to capital income consuming a greater share of national income. This makes a small number of people who own the robots very rich, and leaves everyone else fighting over scraps. Through formal (direct contributions, cash for access) and informal (threat of capital flight, social connections) channels, the rich minority are able to control politics. Inequality driven by automation will enhance their influence and undermine democracy, leaving us with some form of rule by the wealthy. The best way to avoid this outcome is to diminish the power of the rich by establishing public ownership of a substantial chunk of the assets in the economy and distributing the returns equally.