I watched a great talk from Andrew McAfee and Erik Brynjolfsson this week. For a few years now these two have been evangelizing about the coming automation revolution that now seems to be given to wider discussion if not yet acceptance.
In their second book, The Second Machine Age, Brynjolfsson and McAfee return to their diagnosis of extreme economic disruption due to increasing automation in the manufacturing and service industries. I have yet to pick up the book myself, but after watching their Google talk I am sure to do so in the very near future.
In their presentation, they reference some surprising statistics, including that the number of people employed by the manufacturing industry in China has actually decreased in recent years. As automation starts to really accelerate in the manufacturing sector, this might mean that developing economies potentially have much higher exposure to associated economic disruption. What it will mean for the steady climb of the world’s poor if they lose access to the lower rungs of the economic ladder remains to be seen, but it seems entirely possible that automation could lead to serious problems for places like Bangladesh.
The most fascinating thing that I think the two brought up was a very apt comparison of human workers to horses. A hundred years ago, a horse was a very important economic unit. In 1900, if I offered to lease my horse to almost any business owner for a reasonable price, they would have likely taken me up on the deal because they could do a lot of stuff with a horse. They could transport themselves to work, they could deliver goods to market, they could plow a field, or they could sub-lease the horse out to do any of these jobs for other people. Horses used to be useful and profitable.
But if you were to take your horse to most any business person today, even if you offered the horse to them for free, you would probably get laughed out of the building. What the hell am I supposed to do with a horse? This would almost certainly be their response for the simple reason that horses are no longer economically useful. Machines have replaced horses so perfectly that the cost to train, feed and house a horse greatly outweighs their economic benefit. Outside of a few niche nostalgic industries, horses are economically obsolete.
So this of course begs the question, are human workers going to become like the horse? Of course there will be an economically productive place for exceptional people for the foreseeable future (just as there is for exceptional horses), but the question is really about the average worker. Will we soon get to a point where automation is so cheap and so effective that the cost to train, feed and house an average worker will outweigh the economic benefit that can be derived from them.
McAfee says that he and Brynjolfsson have had heated discussions about this very issue, but generally they are still not sure whether we are headed to a future where average workers are economically obsolete. To me it is scary how clear the conclusion is. Yes, human workers are just like horses, we are already on the arc of economic obsolescence and we have been for some time.
There is a simple example that makes me so sure that average workers are slowly falling behind their economic benefit — children. Children used to be a clear economic benefit to their parents. They could be easily put to work doing any of the countless menial tasks associated with an agricultural economy, such as plowing fields, harvesting crops, milking cows and churning butter. Children were economically beneficial, and people responded to this by having lots of them.
But, as the first industrial revolution started making it easier and cheaper to use machines to produce food, a huge number of workers migrated to the city looking for work. In the cities, children lacked the strength, fine motor skills and education that was necessary for them to be really useful in the factory economy. Suddenly, children went from an economic benefit to an economic liability. The cost to train, feed and house a child greatly outweighed their economic benefit, and parents responded by having far fewer of them.
People responding to this change from children being an economic benefit to a cost is a story that has played out over and over again across the world. The economic disincentive to child-bearing has been the single most important factor in the falling fertility rates across the globe. This change is so complete that we are now in a world where the overall average number of children per woman is only 2.5 (see this excellent talk by Hans Rosling and the BBC).
It is a striking fact that the transition of human children to economic obsolescence has been so complete the worldwide birthrate already stands barely above replacement numbers. And in more recent times, I think obsolescence extends well beyond just children. Only a couple of decades ago, an 18-year-old with only a basic education was still of fair economic value. They could get a job in a small business or in a factory and likely be of good economic value to the business owner; at least their value was more than enough for them to survive on. Nowadays this no longer seems to be true, a point reflected in the fact that minimum wage generally falls close to or below the amount of money which is really needed to train, feed and house an average person. We already live in a world where the economic benefit of an average untrained individual is falling behind their maintenance costs.
The creeping arc of economic obsolescence may not stop at uneducated 18-year-olds either. The economic downturn hit young people disproportionately harder than the older population of established workers, and it seems that for youth around the world the recession is not yet over. So many grads, even with good post-secondary educations are unable to find productive employment, and are increasingly showing their agitation at this fact. Economists point to complex macroeconomic changes as the reason behind inflated youth unemployment, but maybe it is simpler than that. Perhaps on the wider scale, the costs of hiring new employees simply outweighs their economic benefit; maybe we are just experiencing the early symptoms of the arc of economic obsolescence reaching the level of the average educated worker.
Of course businesses need to hire some new people, but it only takes a small shift in the balance between the available pool of workers and the number of jobs openings to remove the pressure to offer competitive wages and benefits. All of this, and the automation revolution has not yet even started. Where are we going to be in five to ten years when every call center is automated, cars can drive themselves, and computers are consulting on medical disagnosis?
At the end of their talk, Brynjolfsson and McAfee offer some possible prescriptions for the automation revolution that they predict will hit over the next decade. They offer options such as a negative income tax, which would subsidize workers at the bottom of the pay scale who offer the least economic value. While the institution of a negative income tax or a basic income (a better option in my opinion) might go a long way to easing the effects of high structural unemployment, the problem with this might be summed up by a criticism McAfee himself delivers: What if this is a linear solution to an exponential problem?
When the horses went from benefit to cost, I am sure that the transition was not nice for those who worked in the industries which supported horses (let alone for the horses themselves). I am not optimistic that the transition of the average human worker from economic benefit to cost is going to be any nicer. As our baked-in beliefs about the value of work are increasingly at odds with our economic realities, I think things are going to get much worse before they get better. The social support systems that we now have in place, such as welfare, pensions and unemployment insurance are far to weak to deal with the extreme economic disruption that will be delivered by automation. Add to this the fact that much of the world has no such systems at all, and you can see how precarious our situation is.
It may not be today, and it may not be tomorrow, but sometime soon we will be facing unacceptable rates of unemployment, knock-on damage to the consumer economy, and significant outrage like that already happening in Spain and Greece. People will not wake up and realize how completely the game has changed until they can hear the noise. Put simply, whether we are talking about humans or horses the economic system places no intrinsic value in individuals beyond what work they can do. If a machine can do what you do cheaper and more efficiently, then it will.
Still, I must end by saying that we should not start smashing the looms; the great economic benefit of automation will ultimately outweigh its costs. Automation is going to offer untold economic surplus, but we must take that surplus and turn it towards creating a more humane system. Unless we find a means to institutionalize a basic humanity into our economic systems we will find that more and more of us are just being put out to pasture.