One of the big themes of 2016 has been an increasing concern about automation – the idea that robots are going to steal all our jobs.
So to put minds to rest this festive season, let me say that, yes, robots are indeed going to steal all our jobs. And it’s going to be absolutely great.
We’ve all seen the reports. Oxford academics say that 47 per cent of jobs are at risk of automation within the next few decades. Other researchers claim that it’s robots, not the Chinese, who have wiped out American manufacturing. And now they’re coming after our middle-class jobs too, the indoor stuff with no heavy lifting.
But what isn’t being properly spelt here is how technological change happens – and how jobs are destroyed and created.
The first and most important thing to grasp the importance of productivity. As Paul Krugman has said, productivity isn’t everything, but in the long run it’s pretty much everything .
As and when the economy becomes more productive, then we all, in aggregate, become richer. We can see this from GDP. You can calculate this either as the sum total of all income in an economy, all production, or all consumption – because the three end up being the same.
That means that if production goes up, via greater productivity, then obviously someone’s income must also go up and so also someone’s consumption. It simply is not possible for this not to be true.
In other words, if machines produce more then we must, in aggregate, be getting richer.
Second, it is not true that there is some limited number of jobs available or to do. We take it as axiomatic that human desires and wants are unlimited – indeed, the subject of economics is how to assuage or sate as many of those as possible with the limited resources the universe has made available to us.
One of those scarce resources is human labour. But adding machines to do some of that work doesn’t mean there is nothing for labour to do. It just means that labour moves on to sating some other human desire or want, leaving the previous work to the machines. We cannot run out of work while there are still things which humans want – it is simply impossible.
Of course, if it turns out that human desires are not unlimited, then it is possible to run out of things wanted. But in that case, the limitation is that there are no jobs once every human has everything their heart desires. At which point, who cares about jobs?
At this point, many people put forward a variation on this argument. In this new age of automation, the machines will be doing all the work and the owners of the machines will get all the profits and the rest of us will be left wailing in the wilderness: no work, no income, nothing.
Which is, again, a situation which is impossible. We work to produce things which others consume, and we in turn consume what others produce. If we can consume what the machines produce, then we can just move on to doing some other work, as above. We’re richer by that new production – meaning we can consume over and above what we used to produce. Either we get to consume the machines’ output, in which case we’re richer. Or its owners keep it for themselves, in which case we’re just as well or badly off as we are today.
All of which leaves us with only one real problem – how fast is this going to happen? Too fast would indeed be a problem, as Keynes pointed out.
We are suffering, not from the rheumatics of old age, but from the growing-pains of over-rapid changes, from the painfulness of readjustment between one economic period and another. The increase of technical efficiency has been taking place faster than we can deal with the problem of labour absorption.
Note that that’s from 1930: this is not about the Depression but the generally miserable conditions in the UK in the late 1920s, caused largely (but not entirely) by the mechanisation of agriculture plus the move from steam power to electricity. Plus, of course, Churchill’s idiot decision to rejoin the gold standard in 1925 at the 1914 rate.
Technological change that goes too fast can definitely be a problem – as many a redundant coal miner can tell us all. So we need to consider the speed at which all of this is to happen.
Let’s take that 47 per cent estimate. To try to measure what that means, we need to understand how quickly the economy destroys and creates jobs in normal times, without the algorithms and robots breathing down our necks.
The answer is: a lot faster than anyone thinks. Every year, some 10 per cent of jobs are destroyed –through bankruptcies, downsizing, firms getting out of a particular line of business and so on. But another 10 per cent are created, as firms start up new lines of business or new companies are founded. In fact, this closing of old firms and opening of new ones is one of the main routes of technological advance, as firms built around using the latest tools displace those that aren’t.
As well as the flow of people into and out of the workforce, there is another flow – of people voluntarily leaving one job to go to another. This represents another 10 per cent of all jobs or so each year.
All this suggests that the economy in general should be quite able to absorb the changes predicted from automation. If we expect 600 per cent of all jobs to change over the next 30 years, adding another 47 per cent can probably be done without too much of a struggle.
Of course, we do need to test this. Do we think that we really can accommodate that rate of job destruction and thus cope with the technological change coming our way? Or will we find ourselves in a situation where destruction outpaces creation – which tends to be what happens in a recession?
As it happens, the US is cheering itself on these days for getting new jobless claims (that is, the number of people registering for unemployment insurance for the first time) down to around 250,000 a week.
The last time this number was consistently at this sort of level was back in the 1970s – and earlier, the non-recessionary parts of the 50s and 60s.
But that in turn is misleading – because the American workforce back then was about half the size it is now. Meaning that the rate of job destruction today is half what it was back then.
Today, in any one week, there’s about a 0.15 per cent chance of any specific job being destroyed. Back then, it was more up at the 0.27 per cent, 0.28 per cent level. And this, let’s remember, was in those allegedly halcyon days of job security, before Ronnie Raygun turned up to kill all that was good and holy about strong unions and shared prosperity.
In other words, if we can get the job destruction and creation rates back up to those extreme limits of the 1960s, then we’ll easily be able to absorb the flow of labour through unemployment as a result of technological change.
Yes, if the robots turn up and nick half the jobs next Tuesday afternoon then we’ve got a problem. But if this all plays out over the next few decades as is predicted, the worst that will happen – if the 1960s are any guide – is that Cliff Richard might become hip again.
Not that any of this should really surprise us. We’ve been destroying jobs by the use of machines since 1750, and we’ve not run out of useful things for people to do as yet.
The important thing to remember in all of this is that what displaced workers do is something different. Without mechanised agriculture, for example, we require 95 per cent of the population to stand around in muddy fields, in order to feed 100 per cent of the people. If we’re doing that, then we cannot have an NHS which employs 10 per cent of the population. The tractor thus makes possible the NHS.
The effect of the robots will be just the same. It will free labour from what it produces now to go and produce something else. And we’ll all be richer from whatever that new thing is. I don’t know what it will be, nor do you, nor does anyone at all. That’s why we have a market economy, not a planned one.
But the only end to this process is when there’s nothing more that any human being wants produced. And wouldn’t that be a tragedy?