Part 3 of 10: Think a machine could never do your job? They’re getting much, much smarter

This is the third in a series of ten posts on the threat to jobs and growth from technology and online distribution, and what we might do about it. 

  • The first part summarises the argument
  • The second introduces the challenge.
  • The third looks at the threat to jobs from automation.
  • The fourth looks at the threat to jobs from online competition.
  • The fifth looks at what economic problems this might cause.
  • The sixth looks at the social and moral problems it might cause.
  • The seventh looks at some of the arguments against a policy response: are we really sure this is a problem? Doesnt technology always create as many jobs as it destroys? Surely there is nothing we can do?
  • The eighth explains why more education and training isn’t the solution to technological un and underemployment.
  • The ninth explains why more self-employment and entrepreneurship isn’t an adequate solution either.
  • The tenth looks at other solutions, and proposes a new one.

Right now unemployment in the UK is relatively low.

But the lesson of history is that as technology gets smarter and cheaper, it takes on work which people used to do.

The waves of automation have always started low, washing away only some of the worst paid and least skilled jobs – the lamplighters and ‘knocker uppers’ of history. But as the tide rises, it begins to wash away the more skilled, better paid jobs too – software replacing accountants, for example; self driving cars, ships and planes replacing taxi drivers, captains and pilots.

k u

A ‘knocker up’ – he knocks on your windows to make sure you get up and out to work on time

Optimists will reply that technology has always created as many jobs as it has destroyed.

It’s an important argument and I look at it in post 7.

In this post, I want to focus on one of the reasons that this time might be different from the past: todays technology is increasingly capable of substituting not just for manual work, but for mental work as well.

Why might automation threaten more jobs now than in the past?

The best known paper on this is by Professor Michael Osborne  and Dr Carl Frey. They argue that in the past, most of the jobs threatened by automation were the ones which involved following rules. The call centre operator who asks you what you want and puts you through to line one or two depending on your answer, for example.

But today’s algorithms can do more than that. They can increasingly recognise patterns. They can ‘see’ and ‘hear.’ They are increasingly dextrous.

This, say Frey and Osborne, enables them to perform a broader scope of manual tasks.

They categorised 702 different kinds of jobs, from recreational therapists and foresters, to music composers and substance abuse counsellors, by how susceptible they are to automation. Their sobering conclusion was that 47% of the jobs in the US economy today could be automated.

The Bank of England took their methodology further. As I mentioned in the last post, it calculates that that would mean 80 million jobs lost in the US and 15 million in the UK – half of the total jobs in the UK economy today*. A third of all the jobs in the UK have a 66% chance of being automated away.

A pharmacist, for example, might well feel safe from automation. But a robot pharmacist is already at work at the University of California, writing prescriptions more accurately than a human.

If you offer financial advice you might feel safe too. But youre competing with automated, algorithm-based portfolio robo-advisers and artificially intelligent asset management, which are already managing $50 billion worth of assets. Barclaysformer chief executive expects half of all the jobs in the UK finance sector to go over the next ten years.

Youre a journalist? Associated Press have already automated much of their reporting of  business results, and more will follow. In a recent survey in Sweden, 37% of readers thought an automated sports report was written by a real journalist. The technology is only going to improve.

You work in retail? The British Retail Consortium expects the sector to shed 37,000 jobs a year to automation.

You’re in construction? Here’s a video of a robot brickie.

You’re a lawyer? You may see many law firm jobs disappear as discovery – the initial work of gathering evidence from other party or parties – is automated.

Youre an accountant? The Bank of England puts the chance of your job surviving in the short to medium termat 5%.

You operate a police patrol? There are now robot public guards in Dubai and San Francisco. This way we can provide better services without hiring more people,” is how Colonel Khalid Nasser Alrazooqui, head of Dubai’s smart unit, puts it.

(Youre a teacher, personal trainer, or priest? Fair enough. Youll be fine).

Here is the Bank of England’s calculation about how vulnerable different types of job are. 

Probability

Source: ‘Labour’s Share,’ Speech by the Chief Economist of the Bank of England

Ever helpful, the BBC have set up a handy tool to tell you how likely your job is to be automated away.

Subsequent research building on Frey and Osborne’s method found that across OECD countries, 57% of jobs could be at risk. And the figures for other countries are much worse: 69% of jobs in India, 77% per cent in China, a mind-boggling 85% in Ethiopia.

In short – this time might be different because many technologies will be deployed at the same time, and many of them will replace brain workers.

There is no precedent for this.

Many people will just feel in their gut that such a collapse in employment is impossible. After all, it is not as if the global economy is about to be blindsided by a systemic flaw that had gone unnoticed before.

I think this was a respectable position until the financial crisis did exactly that.

But automation is not the only threat to jobs. In the next part I’ll look at the threat from online competition.

*And of course, automation is not the only aspect which could affect jobs. A quarter of all jobs in the US, for example, are also offshorable.

Scope 2

Standard

Part 4 of 10: Middle jobs are disappearing. Here’s why online competition might be making it worse

This is the fourth in a series of ten posts on the threat to jobs and growth from technology and online distribution, and what we might do about it. 

  • The first part summarises the argument
  • The second introduces the challenge.
  • The third looks at the threat to jobs from automation.
  • The fourth looks at the threat to jobs from online competition.
  • The fifth looks at what economic problems this might cause.
  • The sixth looks at the social and moral problems it might cause.
  • The seventh looks at some of the arguments against a policy response: are we really sure this is a problem? Doesnt technology always create as many jobs as it destroys? Surely there is nothing we can do?
  • The eighth explains why more education and training isn’t the solution to technological un and underemployment.
  • The ninth explains why more self-employment and entrepreneurship isn’t an adequate solution either.
  • The tenth looks at other solutions, and proposes a new one.

My first job was in a video and DVD rental shop. The guy who started it was dedicated, smart, and hungry for success.

Still, it soon failed.

Of course it did: it was a video rental shop. They all failed.

But they didnt fail because they were replaced by robots or artificial intelligence. They failed because the distribution of films changed: we started getting them online.

This illustrates why I suspect the current way we think about technological unemployment underestimates the problem.

Jobs are not just under threat from technology substituting for humans. They’re also under threat from online competition.

Once competition in an industry moves online, it becomes much flatter, and sometimes global. Margins normally fall. These industries can only sustain revenues and employment if growth in emerging markets offsets decline in mature ones. Otherwise, they will sustain far fewer jobs. This is already the story of newspapers in the UK and the US, recorded music (thanks to Spotify* etc), traditional taxis (Uber), and other industries.

And if competition makes the product free – as it already has with much music and journalism – the industry is able to support even fewer jobs.

This is already the trajectory in academia, with the rise of online courses and peer-to-peer paper sharing.

It’s the trajectory in local markets for household tasks, where sites like Taskrabbit.com and fountain.com offer an online marketplace.

And it’s the trajectory in legal services. Online marketplaces like legalzoom.com makes formerly local markets national; lawyers sometimes find themselves competing with thousands of others all over the country, threatening fees and margins.

But here’s the rub: I expect in the next decade or two the same will happen in the great job creator of the twentieth century: manufacturing.

When 3D printers are cheap enough for the median worker to afford, she will be able to download and print some of the things she used to have to go out and buy or have delivered. At that point, we can expect downward price pressure, shrinking margins, and fewer jobs across manufacturing.

How long will it take before we have a Spotify for 3D printable objects’?

Think for a moment what that will mean. When 3D printer ownership is sufficiently widespread, a ‘Spotify for 3D printable objects’ would mean that you could download and print objects – toys, tools, trinkets, whatever – for free.

That might be a tough idea to take seriously. But when we all bought music on CD, it was tough for the music industry to take the threat of free music seriously too. 

If a ‘Spotify for 3D printable objects’ happens, manufacturing jobs are likely to crater.

It probably also means fewer distribution jobs too. Once you can download and print out items at home, fewer goods will need transporting. Jobs are likely to go in parcel delivery, retail and wholesale, and transportation systems, like trucks and container ships (if they haven’t already been automated anyway).

In other words, one lesson from music, taxis, and newspapers is that even if robots dont eat your job, global online competition and distribution can still shrink your industry.

We will have a race between the broadly ‘job-creating’ forces and the broadly ‘job destroying forces.

On the job creating side we will have the rising middle class of developing countries providing the world economy with more demand, and the income effect: cheaper goods and services saving consumers money so they have more to spend on other things, creating new demand and jobs.

But on the job destroying side we will have automation and artificial intelligence requiring fewer humans to produce the same output, and online competition pushing down margins in new and existing industries, putting downward pressure on total demand and jobs.

It’s too early to say which will win between these forces.

But I wouldn’t bet on the job-creating forces. Not least because so many products online are infinitely replicable (music, books, journalism, products when 3D printing becomes ubiquitous). So however many Brazilians, Indians, Nigerians and others emerge into the middle class, as soon as they buy online, you surely do not create as many jobs as you would have done before the internet was around.

Some will feel instinctively that this world I have sketched sounds great – its main feature being cheap goods and services.

But these trends might cause real problems. In the next post, Ill look at why.

*Some will point out that Spotify has actually helped the music industry recapture revenues they were losing to pirating. True, compared to ten years ago. But compared to twenty, thirty, or forty years ago, the most important thing Spotify and YouTube have done is radically reduce the value of recorded music, and helped to make the music industry much less profitable.

14128390_s

Standard

Part 5 of 10: Here’s why fewer middle jobs is bad news for everyone.

This is the fifth in a series of ten posts on the threat to jobs and growth from technology and online distribution, and what we might do about it. 

  • The first part summarises the argument
  • The second introduces the challenge.
  • The third looks at the threat to jobs from automation.
  • The fourth looks at the threat to jobs from online competition.
  • The fifth looks at what economic problems this might cause.
  • The sixth looks at the social and moral problems it might cause.
  • The seventh looks at some of the arguments against a policy response: are we really sure this is a problem? Doesnt technology always create as many jobs as it destroys? Surely there is nothing we can do?
  • The eighth explains why more education and training isn’t the solution to technological un and underemployment.
  • The ninth explains why more self-employment and entrepreneurship isn’t an adequate solution either.
  • The tenth looks at other solutions, and proposes a new one.

In the last two posts I sketched why automation and online competition could drastically reduce the number of middle jobs and the total number of jobs.

Here’s the first of two posts on why that would be a problem.

1. Lower Economic Growth

Fewer jobs – or even just fewer middle jobs – will mean less purchasing power across the economy.

Jobs are simply the main way that money flows from companies to the broad mass of people who consume. When there is a shortfall of jobs, there is a shortfall of purchasing power and demand, holding back growth.

This sometimes happens already, of course. In 2014, for example, the International Labour Organisation joined many commentators in worrying about low aggregate demand and slow job creation across G20 countries.

If middle jobs carry on disappearing, this will be an issue because the rich spend less of their income than the rest. Fewer middle jobs, let alone higher structural unemployment, would be bad for business: companies would be fishing for business in an ever smaller pool of purchasing power.

Ultimately, it would surely threaten long-term trend growth.

This may already be happening; long-term growth in developed countries seems to have stalled (‘secular stagnation’), and economists are debating why.

2. Deflation

Fewer middle jobs could lead to deflationary pressure, and perhaps even deflation.

If enough jobs are lost or pay falls, total spending power in the economy falls, so companies might be inhibited from putting up prices. They might even decide they need to drop them to compete.

Sustained, across the board deflation offers something for everyone not to like.

Wages would fall. People might put off purchases while they wait for prices to fall further, like you might do before a sale that you knew was due to start next week. That make it tough to do business. Debts would rise in real terms each year. And businesses would be discouraged from borrowing to invest, because they would have to pay back more than they borrowed in real terms.

This is the story of the USA during the Great Depression.

On a small scale, we already have historically high deflationary pressure.

Andy Haldane, the Bank of Englands Chief Economist – has said (p.17) that the monetary policy committee has been consistently surprisedby how weak wage growth is in the UK.  And it could be because of the effect of technology on jobs. As he speculated, it could be because technology has made it easier and cheaper than ever before to substitute labour for capital, man for machine.

3. A waste of education and skills

A loss of middle jobs would be a waste of education, skill, and human potential.

It would mean more people with hard-won skills doing jobs which dont use those skills, like the guy with two degrees and a business school Masters cleaning the toilets in Pret, or the rise of the PhD barista.

Over time, the problem would probably get worse. After all, the demand for higher education probably wouldnt fall substantially – it would still be the best chance to get the best jobs. But if the number of graduates grows faster than the number of jobs which need graduate skills, you are left with more overeducated and overtrained people each year. Thats when the skilled people start leaving the country.

Again, there is some evidence that this is already happening: in the first twelve years of the century, ‘skill unemployment’ doubled in the US and the UK and tripled in Italy and Spain. 

Could it be because in those countries, ever more skilled workers are chasing ever fewer jobs which need their skills? This is an indicator to watch.

4. Lower Productivity

Productivity drives growth and wages, so a waste of skills means the economy is less productive than it should be.

That raises issues of national competitiveness.

If, say, France finds a way to protect its middle jobs but Britain doesnt, then French skilled workers use those skills productively but British skilled workers dont. French productivity pulls further ahead of British productivity, and the motor powering the French economy gains power while the British motor sputters.

Substitute France for China, Brazil, or Nigeria – or any countries with a growing middle class of consumers – and the challenge compounds. The Civil Service has already started worrying aloud (p.29) about this.

And, sure enough, British productivity has stalled in recent years. 

VR

Source: ‘Labour’s Share,’ Speech by the Chief Economist of the Bank of England

If we continue to lose middle jobs or all jobs, skills will be wasted, placing a drag on the economy compared to its potential.

5. Lower Interest Rates

A loss of middle jobs or aggregate jobs means it will be harder for people to work their way out of debt.

That could discourage central bankers from raising interest rates, as they might be inhibited from making it harder for debtors to pay their debts down. But those low interest rates would also be a disincentive to save.

It would also make it harder for central bankers to use interest rates to stabilise the economy. If central banks consistently expected low demand and inflation, they would be obliged to keep interest rates low, blunting one of the most useful tools by which policymakers try to regulate the economy.

This last aspect has already happened, of course. And at least one member of the Monetary Policy Committee has explicitly linked the problem to distributional issues – inequality.

This is a sample of the economic problems which higher technological unemployment or underemployment could cause.

The next post looks at three social and moral problems.

13771502_s

Standard

Part 6 of 10: Fewer middle jobs isn’t just an economic problem. It’s a moral problem too.

This is the sixth in a series of ten posts on the threat to jobs and growth from technology and online distribution, and what we might do about it. 

  • The first part summarises the argument
  • The second introduces the challenge.
  • The third looks at the threat to jobs from automation.
  • The fourth looks at the threat to jobs from online competition.
  • The fifth looks at what economic problems this might cause.
  • The sixth looks at the social and moral problems it might cause.
  • The seventh looks at some of the arguments against a policy response: are we really sure this is a problem? Doesnt technology always create as many jobs as it destroys? Surely there is nothing we can do?
  • The eighth explains why more education and training isn’t the solution to technological un and underemployment.
  • The ninth explains why more self-employment and entrepreneurship isn’t an adequate solution either.
  • The tenth looks at other solutions, and proposes a new one.

A structural loss of middle jobs or total jobs wouldn’t just have economic consequences. It would have social and moral consequences too. Here are three.

6. Worse Health

There is plenty of evidence that being stuck without work, enough work, or the right kind of work is bad for your health.

A young person without a job, for example, is more likely to have mental health problems in their 40s or 50s, let alone a loss of confidence, cognitive skills, and optimism. Even just being underemployed or being overqualified for the job can have the same effects. Of course, this increases the burden on the health service and everyone who works in it.

7. Inequality

Jobs are the main way money flows from the owners of companies to everyone else.

When economies work like they did in the US and Western Europe in the middle of the twentieth century, money flows from the owners of companies to the majority of people, who in turn buy those companies’ products. It’s a virtuous circle.

But if there are fewer jobs or middle jobs, companies can grow but the money flows into the pockets of a smaller and smaller pool of employees. That means customers are poorer than they would be.

John Lanchester puts it vividly:

Imagine an economy in which the 0.1 per cent own the machines, the rest of the 1 per cent manage their operation, and the 99 per cent either do the remaining scraps of unautomatable work, or are unemployed. That is the world implied by developments in productivity and automation. It is Pikettyworld, in which capital is increasingly triumphant over labour.

The process has started. We can’t be sure technology is the cause, but for the first time since the war, the share of US GDP that goes to owners (capital) has begun to definitively pull away from the share going to workers (labour).

Erik Brynjolfsson’s chart (from this presentation) shows how that looks in the USA.

The blue line shows corporate profits after tax, and the red line shows the non-farm labour share, both as a percentage of GDP. They skip around each other from 1947 to 2002; sometimes labour has more, sometimes capital. But from 2002, the difference is dramatic. They diverge sharply; owners make ever more, workers make ever less.

c & l

That’s also the pattern in the UK and across rich and middle income countries.

ILO2

This chart comes from the ILO’s report on G20 Labour Markets, p.13

Many political centrists are beginning to wake up to this. Centrists in Britain used to believe that a rising tide lifted all boats. Not any more. Many, like John Major and Tony Blair have accepted that inequality is a problem. 

Recent profit margins have even had one analyst at Goldman Sachs wondering about “the efficacy of capitalism.”

David Autor, economics professor at MIT and a leading thinker on automation and employment, sees these distributional issues as the main challenge of technology.

8. Effort and reward

We like to believe that the harder you work, the better you do in life.

Its what we teach children. Politicians and brands are keen to associate themselves with the sentiment. Social media is awash with pithy posters expressing it. We are culturally accustomed to being prouder of what we earned than what we were given. Nobody who inherits a house boasts about being a homeowner.

We like to think that this morality tale applies to economic life too, that the rich are rich because they have worked hard, that they deserve it. Sometimes its true.

But if the number of middle jobs declines, then there will simply be fewer middle jobs for graduates to get. People will still put in the same effort to get them – they might even work harder – but fewer will succeed, because many of the middle rungs of the ladder which were there for the last generation won’t be there for the next.

That probably means reduced equality of opportunity as well. Like it or not, money buys opportunities for the next generation: tutors, remedial classes, networks.

Unless you have reason to think current trends towards the hollowing labour market will reverse, fewer middle jobs means fewer households able to pass these opportunities on to their children.

In short, if automation and online competition continue to eat into middle jobs – as they have been doing for years in the US – we can expect lower purchasing power and growth, deflation, wasted education and skills, lower productivity, lower interest rates, worse health outcomes, increasing inequality, and the decoupling of effort and reward, as each decade brings greater accumulations of wealth for the richest and greater debt for the poorest.

But should we do anything about this? Could we? These are the questions I will look at in part 7.

13766384_s

Standard

Part 7 of 10: Some people say we shouldn’t do anything about technology taking jobs. Here’s the counterargument.

This is the seventh in a series of ten posts on the threat to jobs and growth from technology and online distribution, and what we might do about it. 

  • The first part summarises the argument
  • The second introduces the challenge.
  • The third looks at the threat to jobs from automation.
  • The fourth looks at the threat to jobs from online competition.
  • The fifth looks at what economic problems this might cause.
  • The sixth looks at the social and moral problems it might cause.
  • The seventh looks at some of the arguments against a policy response: are we really sure this is a problem? Doesnt technology always create as many jobs as it destroys? Surely there is nothing we can do?
  • The eighth explains why more education and training isn’t the solution to technological un and underemployment.
  • The ninth explains why more self-employment and entrepreneurship isn’t an adequate solution either.
  • The tenth looks at other solutions, and proposes a new one.

If there is a structural loss of middle jobs or total jobs, should politicians do anything about it?

The prospects of wasted education, worse health outcomes, and increased inequality are enough to horrify someone with social democratic instincts like me. But plenty of people will see those same trends and feel equally strongly that we can’t or shouldnt do anything about it.

And when they convert those feelings into arguments, here are some of the arguments they’ll use. 

1. Were not yet sure its a problem

As Ive set out in previous posts, the data is still coming in.

We know technology destroys middle jobs – American and European labour markets are hollowing out. We can see that rich countries like the US and the UK have growing skill unemployment, and that productivity, investment and median incomes have stalled relative to long term trend.

But will automation and online competition really kill so many jobs? Theres still room for reasonable people to disagree.

And, as a busy politician might add, voters are angry for plenty of reasons, but aggregate effects like ‘a hollowing labour market’ is just not something most people think about.

But this is silly. Such devil-may-care logic doesnt afflict political debate when discussing other risks, such as cybersecurity or nuclear threats. There is enough evidence to believe that we might have the beginnings of a serious problem, and it is politiciansjob to address possibilities as well as certainties.

The debate about the collapse of middle jobs and purchasing power is today in a similar place to where the political debate on climate change was in the early noughties.

Then, credible sources in the scientific community were concerned with it, some in policy circles knew about it, but it was still seen as a fringe risk. Hardly any politicians had started thinking seriously about what we should do.

Even though we arent sure we will face structural un or underemployment, we can be pretty confident that if it happens, the consequences will be dire.

Why wouldn’t we start thinking about it?

2. Technology could create as many jobs as it destroys

Technology doesnt just destroy jobs, it also creates them.

For all the lamplighters and lathmakers who have fallen into history, there are many more digital brand managers, graphic designers, and database managers. There are deodorant testing armpit sniffers. There is Shingy.

Over the last two centuries, jobs have come and gone, but the UK employment rate has fluctuated around a relatively fixed average, as this graph shows.

Dimsdale

Source: Speech by Andy Haldane: ‘Labour’s Share’

It might do in future as well. After all, there is no limit to what humanity might imagine or demand, so no limit to what the economy might have to supply.

No doubt supporters of the UK government would make this kind of argument. ‘Look’, they might say, ‘all this talk of a hollowing labour market is so much carping. UK unemployment just hit a ten year low. The employment rate is at its highest since records began. There will always be enough jobs.

I hope they’re right. 

But employment now only gives us one data point. The trend is more important.

And there are at least five reasons to think that this view about the future is too complacent.

1. As set out in post 2, this time, automation will affect mental work as well as manual work. Machines can increasingly learn, recognise patterns, sense, and be taught dexterity. All this is new in human history.

2. As set out in post 3, this time we are reckoning with online competition, often global, which puts unprecedented downward pressure on prices and margins, making it hard for existing industries to support as many jobs. This too is new in human history.

3. The industries of the future probably won’t create as many jobs as the industries of the past.

Martin Ford puts it well: “you can imagine lots of new industries: nanotechnology and synthetic biology – but they won’t employ many people. They’ll use lots of technology, rely on big computing centres, and be heavily automated.”

You can already see this trend at work. Take photography, for example. As Henrik Killander points out, at its pinnacle, Kodak employed more than 140,000 people; when Instagram was sold to Facebook in 2012, it employed 13.

It’s hard to find research indicating that technology creates lots of jobs, but it’s not hard to find research showing that it doesn’t. One study calculates that of all the new jobs created in the US in 2010, only a puny 0.5% were in industries which didn’t exist at the turn of the century. It looks more and more like new technologies create plenty of wealth but not much work.

I don’t think we yet know enough to say for sure that the new industries of the future will need fewer jobs than the industries of the past.

But one thing is for sure: business leaders look for the most efficient way to get the task done, and labour is a cost. If technology can do a better job than people, or the same job more cheaply than people, or both, they will employ fewer people. Nobody goes into business first and foremost to create jobs. If the industries of the future don’t need as many jobs as those of the past, fewer will be created.

Uber is a salutary example. It talks a good game about how much work it has provided to drivers around the world. But it is still working on replacing them with self-driving cars.

3. Higher unemployment which lasts for five, ten, twenty years is still an apocalyptic problem for the generation which experiences it.

Often people argue about technological unemployment as if it’s something that either will or won’t happen, forever. One side says employment will hold up – technology will create as many jobs as it destroys. The other side says this time it might leave us with fewer jobs permanently. But something else could happen: technology could destroy many jobs in the short run and create just as many in the long run, but give us five, ten, twenty years of transitional high unemployment. 

Long-term, high, but non-permanent unemployment has happened before. Whether or not it was because of technology, unemployment in Britain stayed above 10% from the early twenties until the beginning of the war.

In other words, even if the optimists are right that technology will create as many jobs as it destroys in the long-term, it can still devastate purchasing power, growth, and human potential for a generation.

4. Fewer middle jobs can still hurt demand and growth. The total number of jobs isn’t the only thing that matters. Getting enough money into enough pockets is important too. Even if technology creates as many jobs as it destroys, if the labour market carries on hollowing out – losing middle jobs – we can still expect a net loss of purchasing power, demand and wellbeing: security, status, opportunity, and development, because the rich spend a lower proportion of their income than everyone else.

And for me, this is the strongest counterargument to the boast that today’s low unemployment means there is no problem: boasting about how many jobs have been created but ignoring how much they pay is like boasting how many cans of beer you’ve got in the fridge but ignoring how little beer is left in each can.

Whereas permanent technological unemployment is a prospect, hollowing out is already happening.

I think we can be pretty confident it has depressed demand relative to trend: many more workers are on the minimum wage. Fifteen years ago, only one in fifty jobs were paid minimum wage; today it is one in twenty, and its predicted to be one in nine by 2020.

3. There is nothing we can do

Tyler Cowen, for example, believes that in the coming decades, the top ten to fifteen percent of workers whose skills will complement those of intelligent machines will prosper, and the rest will stagnate. In his telling, it’s inevitable.

A flippant way of putting the argument is: if driverless vehicles are going to kill driving jobs and 3D printing is going to kill factory jobs, then, with the best will in the world, theres not much any Minister for Work and Pensions can do about it.

I don’t buy this.

Market forces are shaped by laws, institutions, customs, and culture. They have, at various points in history, involved owning slaves, employing children, and letting multinationals negotiate their own tax rates. These things are inevitable until politics changes the rules. Not everyone will like the solutions, but that doesnt mean there arent any.

But if technology really does reduce jobs and work, isnt the solution just better education and training, skilling up?

In the next post Ill look at this argument.

13770852_s

Standard

Part 8 of 10: Why more education and training won’t fix technological unemployment

This is the eighth in a series of ten posts on the threat to jobs and growth from technology and online distribution, and what we might do about it. 

  • The first part summarises the argument
  • The second introduces the challenge.
  • The third looks at the threat to jobs from automation.
  • The fourth looks at the threat to jobs from online competition.
  • The fifth looks at what economic problems this might cause.
  • The sixth looks at the social and moral problems it might cause.
  • The seventh looks at some of the arguments against a policy response: are we really sure this is a problem? Doesnt technology always create as many jobs as it destroys? Surely there is nothing we can do?
  • The eighth explains why more education and training isn’t the solution to technological un and underemployment.
  • The ninth explains why more self-employment and entrepreneurship isn’t an adequate solution either.
  • The tenth looks at other solutions, and proposes a new one.

If technology really does reduce jobs and work, isnt the solution just better education and training, skilling up?

This seems to be the conventional wisdom.

This solution comes in a number of forms: more money for schools, colleges, and universities, more and better apprenticeships and training, on or off the job. Perhaps an expanded scheme of career development loans and lifelong learning schemes.

When the Civil Service considers more automation and unemployment, this is what it recommends, for example here (p.21).

More and better education is a good thing in itself. But it wont fix the problem.

A good education, training scheme or apprenticeship will only mean a good job if there are enough good jobs out there.

If the problem is not enough middle jobs, the solution is more middle jobs. If the problem is not enough jobs, the solution is more jobs.

Larry Summers put it well:

If we allow the idea to take hold that all we need to do is: there are all these jobs with skills and if we can just train people a bit, then theyll be able to get into them and the whole problem will go away. I think that is fundamentally an evasion of a profound social challenge. The core problem is that there arent enough jobs.

That challenges a very ingrained mindset.

Parents are accustomed to waving their children off to schools or universities safe in the belief that their education means they can expect a good job. And yes, it will always help. But the bigger the hole where the middle jobs should be, the harder it will be for education alone to fix the problem.

And in one sense, more education and training is not helpful because it adds to the demand for middle jobs, and so bids down their salaries. Italy, Spain, the US, and the UK have all experienced rising skill unemployment because they have ever more skilled graduates chasing a dwindling number of skilled jobs. Germany hasn’t. As economist Dalia Marin put it: ‘in Germany, skill unemployment is low and did not increase between 2000 and 2012 precisely because education was advancing slowly there’.

This boils down to an uncomfortable trade off. In a country with a permanently lower proportion of middle skill jobs, should we prefer commensurately fewer skilled people to do them? Or an equal or greater proportion of skilled people, even if it means more people get stuck in jobs they’re overqualified for?

Anyway, if education and skills arent the answer, what about entrepreneurship? Surely if there are fewer jobs, more people are just going to have to create their own?

The next post looks at this argument.

13696968_s

Standard

Part 9 of 10: Why ‘more entrepreneurship’ isn’t the solution to technological unemployment

This is the ninth in a series of ten posts on the threat to jobs and growth from technology and online distribution, and what we might do about it. 

  • The first part summarises the argument
  • The second introduces the challenge.
  • The third looks at the threat to jobs from automation.
  • The fourth looks at the threat to jobs from online competition.
  • The fifth looks at what economic problems this might cause.
  • The sixth looks at the social and moral problems it might cause.
  • The seventh looks at some of the arguments against a policy response: are we really sure this is a problem? Doesnt technology always create as many jobs as it destroys? Surely there is nothing we can do?
  • The eighth explains why more education and training isn’t the solution to technological un and underemployment.
  • The ninth explains why more self-employment and entrepreneurship isn’t an adequate solution either.
  • The tenth looks at other solutions, and proposes a new one.

I think my dad is worried about how worried I am about the future of work.

The other day he emailed me an article about the rise of self-employment in the EU, with the message: you could stop worrying about the future of work, as so many people are finding their own solutions!

And he is well qualified to say so; he started a small business in 1987 and has run it ever since.

Certainly, anyone who has created their own job or dreams of doing so probably cringes at the idea that the total number of jobs can fall. After all, if you have defined your working identity by making a job not taking a job, cant everyone? Isnt the solution to fewer aggregate jobs more self-employment, entrepreneurship, and enterprise?

The argument acquires a moral flavour if you add words like self-relianceand rugged individualism.’ If robots, technology, or online competition mean fewer jobs, wont people just have to pull up their socks and take some personal responsibility?

The argument also has a strong cultural aspect. We want to believe this. If an American ‘single mom’ working low-paid jobs with three kids can make it big with a mop head you can wring out without getting your hands wet, cant anyone?

I would never discourage anyone from trying.

But that doesnt mean the economy can automatically expect to rely on self-starters to keep the jobs numbers and aggregate demand stable, for three reasons.

1. Most self-employed people dont make much money

In 2013-14 in the UK*, the typical self-employed person made roughly half the median wage of employees. (£210 per week (p.21) vs £385 per week).

And 2013-14 wasnt a one off. Since the turn of the century, the typical self-employed person has earned between two thirds and half of the median employee, as this graph shows.

The dotted lines – the medians – are the relevant ones here because the full lines – the means – are skewed by a few high earning self-employed contractors.

The red dotted line is median self-employed earnings, the green is employee earnings.

ResFound2

Source: The Resolution Foundation: ‘All Accounted For: the case for an ‘all worker’ earnings measure’

2. If jobs move from big companies to small ones, there would still be fewer middle jobs

Okay, you might say, so self-employed people don’t tend to make as much money as employees. But what about new companies which grow and employ new people? If middle jobs carry on disappearing, couldn’t new companies replace them?

I hope so. But it’s a risky bet. The conventional wisdom is that growth is driven by improvements in productivity, improvements in output per hour of work. Traditionally, bigger companies have been more productive than smaller ones; they have economies of scale and it’s easier for them to afford new technologies.

You can already get a sense of that when you look at the businesses in the UK today.

In the UK, most businesses are small, but the businesses with the most turnover are big. 99.9% of businesses employ under 250 people, but 53% of all the turnover in the country was made by the remaining 0.1% – the big boys.

Unfortunately, all other things being equal, that means that an economy-wide shift from employment in big companies to employment in small ones means an economy-wide shift from companies which make a lot of money to companies which don’t make nearly as much, from more productive big companies to less productive small ones.

That would make it harder for an economy of small companies to pay as much as an economy of big companies, putting downward pressure on pay, demand, and growth.

3. Self-starters would have to replace a lot of jobs

The Bank of England estimates that automation could threaten 15 million UK jobs in the next decade or two. If that turns out to be right, would self-starters be able to replace those jobs?

Maybe, but it would be hard.

Here’s a very rough and ready way of thinking about it. Britain, with its relatively entrepreneurial culture and ease of doing business, has 12.4 million people working in small businesses (firms with fewer than 50 employees).

Assume for the moment that all of the jobs at SMEs – companies with 250 or fewer employees – were safe. In other words, that the 15 million jobs lost to automation came from big companies with deep enough pockets to take on the labour-displacing technology first. Let’s also assume that the workforce stays the same size: perhaps net migration cancels out the effects of an ageing population.

In that scenario, the only way the employment rate could stay steady would be if self-starters or small companies created 15 million new jobs. That’s more than the 12.4 million they have created to date. That’s a big ask.

In reality, plenty of other things would complicate the picture. Perhaps the Bank’s estimate for jobs lost to automation could prove optimistic or pessimistic. Perhaps automation eats into the numbers employed by small companies as much as big companies. But in that case, self-starters would have to create even more jobs to keep the employment rate steady.

But I think we can be sure about two things.

Historically, most of the countries which have enabled the broad mass of people to get richer together are those which have produced a mass of middle-income jobs, like Germany or France in the decades after the war. One of the reasons poor countries are poor is because they tend to have a lot of self-employed people. I think about the people selling fruit, snacks, and trinkets in the streets. These jobs are better than nothing, but they’re not a strategy to bring a broad mass of people towards prosperity.

Secondly, if millions of people do move from being employees to self-starters, it would represent millions of self-starters who would prefer a job – a seismic cultural shift.

It’s not hard to see why most people who come out of education want a job: they want the certainty of regular pay, knowing when they can expect to pay off their student loan, build a credit history, get a mortgage, learn skills, not to mention security, status, structure, and the smaller things you take for granted when you’re an employee, like knowing there will be a Christmas party whether or not you organise it.

In short, even though an increase in self-employment and entrepreneurship would be good news, we shouldn’t expect it to maintain the employment rate, purchasing power, or demand.

But if more education and training or self-employment wont solve the problem of lower growth due to a big loss of middle jobs – what might? In part 10 Ill look at that.

*All these figures are British. But compared to most other European countries, Britain has a relatively good culture and environment for self-employment, so the argument is likely to hold for most countries.

34404877_s

Standard

Part 10 of 10: If companies won’t create enough jobs, government should

This is the last in a series of ten posts on the threat to jobs and growth from technology and online distribution, and what we might do about it. 

  • The first part summarises the argument
  • The second introduces the challenge.
  • The third looks at the threat to jobs from automation.
  • The fourth looks at the threat to jobs from online competition.
  • The fifth looks at what economic problems this might cause.
  • The sixth looks at the social and moral problems it might cause.
  • The seventh looks at some of the arguments against a policy response: are we really sure this is a problem? Doesnt technology always create as many jobs as it destroys? Surely there is nothing we can do?
  • The eighth explains why more education and training isn’t the solution to technological un and underemployment.
  • The ninth explains why more self-employment and entrepreneurship isn’t an adequate solution either.
  • The tenth looks at other solutions, and proposes a new one.

The story so far: economists are beginning to think seriously about the possibility of technological unemployment – automation making an unprecedented number of jobs unnecessary.

Frey and Osborne calculate that 57% of the jobs in OECD countries are at risk. The Bank of England conclude that that would be equivalent to 15 million jobs in Britain and 80 million in the US.

Historically, technology has created as many jobs as it has destroyed. But I’m not sure it will this time, as I explained here.

The debate is beginning to move towards a discussion of solutions.

In moments of acute crisis, radical ideas suddenly dont look so radical any more.

If in 2007, Gordon Brown had proposed allocating 80% of the governments annual spending to private banks in loans and guarantees, everyone would have said he was bonkers. The year after, he did just that to stave off the financial crisis. The move received widespread support. The prospect of financial meltdown made the radical sensible.

If it happens, widespread technological un and underemployment would be a chronic problem, not an acute one; there would probably never be a crunch moment when it was clear how radical the solution might have to be.

Still, thoughtful centrists are beginning to conclude that if unemployment does begin to creep up to levels not seen in generations – say 15%, 20% – or the labour market hollows out in an extreme way, there is no private sector solution, and probably no solution that falls within the bounds of what is electable today.

Former US Treasury Secretary Larry Summers, for example, says that in any solution, the tax and transfer system has got to be a very, very large part of the picture.

Robert Rubin, deregulating Treasury Secretary and Goldman Sachs veteran, wondered whether, if the forces of technology and globalisation continue to create rising inequality[there should be] increased redistribution to accomplish the broad objectives of our society?

Various solutions have been proposed.

The Directors of the McKinsey Global Institute suggest the marketisation of household work such as cooking, cleaning, and childcare.

Citibanks Chief Economist Willelm Buiter and Martin Ford suggest a minimum income for all.

There seems to be a consensus that the only policies which might work are radical ones.

I agree. If unemployment reaches unprecedented levels and stays there for a decade or more, one solution would be to create an enlarged sovereign wealth fund and use the dividends to create jobs.

It makes particular sense for the fund to buy shares in technology companies like Alphabet where profits are exceptional and job creation is low compared to the big companies of the past.

The Shareholder State

If millions of jobs are lost and not replaced over the coming decade, it will in part be because companies, understandably, see opportunities to make more revenue more efficiently from automation.

In this world, as long as demand holds, some firms, especially technology firms, would be extremely profitable. Already, six of the top ten biggest companies by market cap on US stock exchanges are tech companies.

This wealth is already taxed – a bit. But if  in the long term we find ourselves in a world of unprecedented long-term structural unemployment, states could also raise revenues by buying shareholdings in the companies which benefit the most from the automation and flat competition which displace workers. These shareholdings would form part of a sovereign wealth fund.

To an extent, this already happens.

In France, for example, the government manages over €100 billion worth of shares in over 70 French companies, including many of the most profitable and innovative companies in the country.

Alaska decided in the seventies to share out oil profits equally to each citizen. Each year since 1982, each Alaskan has received a deposit – sometimes of a couple of thousand dollars – into their bank account. 

There are already 73 sovereign wealth funds worldwide, many non-oil and gas. They are a good idea for many other reasons anyway. 

Renewable energy is also a particularly apt sector for state shareholding. The case that resources like the sun and the wind belong to every citizen is easy to make. As the Alaskan example shows, the decision on who is entitled to benefit from energy companiesrevenue is, ultimately, political.

Energy companies will counter that they take the risk so they deserve the revenue. But this is a norm, not a natural law. Statoil, the Norwegian oil and gas multinational, takes risk too, but it is still 67% owned by the Norwegian government, and contributes to the Norwegian sovereign wealth fund.

This sovereign wealth fund would be a means to an end of supporting demand and growth by creating jobs, by spending the dividends on job creation programmes. 

A New Deal for Jobs

The best solution to a loss of jobs is creating new jobs.

Thats what the US did when it was facing unprecedented loss of jobs after the crisis: the 2009 Recovery Act created or saved 1.6 million jobs a year for four years.

It spent $279 billion* putting people to work mending bridges, roads, and railways; building new trains and rail routes; cleaning property; retrofitting diesel engines to reduce their carbon dioxide output; making tap water safer to drink; improving schools; researching energy efficiency which led to progress on biofuels, more efficient batteries, superconducting wires and vehicles powered by natural gas. It paid the salaries of people working on wind, solar, and geothermal energy; improving hospitalsand surgeriesaccess to health information; building tens of thousands of miles of broadband lines, and training millions of people how to use it.

But the USA was not exceptional. Plenty of countries around the world also stimulated their economies, even if most didnt make direct job creation their primary objective. Economists now have plenty of evidence about what does and doesnt work.

Granted, dividends from a sovereign wealth fund would be unlikely to generate enough cash to fund public jobs on the scale the Bank of England research anticipates. Granted, there would be managerial hurdles. But the point here is to raise the idea.

The jobs which a new New Deal should create would depend on the political priorities at the time.

If it were being enacted in Britain today, I would suggest installing more residential and commercial renewable energy sources, training more carers to prepare for an ageing population, reducing waste and improving sustainability, shoring up flood defences, and building more houses**.

But my preferences arent the point. The point is we do not lack for work to be done.

Conclusion/TLDR

Technology and online competition could mean a net loss of millions of jobs and an unprecedented structural hit to purchasing power, demand, and growth.

The most authoritative research by Frey and Osborne calculates that 57% of the jobs in OECD countries could be at risk. The Bank of England concludes that that would be equivalent to losing roughly half the jobs in Britain today.

Its too early to say whether or not the worst case scenario will happen. We will have a race between the broadly ‘job-creating’ forces – the rising middle class in developing countries and cheaper stuff saving consumers money – and the ‘job destroying forces’ – automation and artificial intelligence and online competition pushing down margins.

But it’s not too early to say that we are heading for fewer middle jobs and more skilled workers doing less skilled jobs. Technological underemployment is with us today.

And surely this has contributed to some of what we’ve seen for the last few years: below trend investment and demand in the US and the UK, rising skill unemployment in some countries, and growth rising much faster for the richest than the majority.

Jobs are the main way money flows into the pockets of the vast majority of people.

If in the next decade or so we do get unprecedented technological unemployment, the consequences would be: lower purchasing power and growth, deflation, lower productivity, lower interest rates, worse health outcomes, increasing inequality, the decoupling of effort and reward, a pool of wasted education and skills, ever greater accumulations of wealth in the hands of the owners of the technologies and greater debt at the bottom.

It is sensible to start to think about how to plan for this possibility. Already, low long-term growth has led to calls for developed countries to take advantage of record low, in some cases negative, interest rates to borrow and spend more to stimulate their economies, such as this one from the OECD.

If it becomes clear that large scale unemployment and underemployment is dragging down demand and growth, the best solution is to create new jobs. As in the response to the global financial crisis, if companies wont, governments should.

One solution is a permanent new deal to support employment and growth, funded from a sovereign wealth fund, comprising shareholdings in the companies which benefit the most from the automation and online competition which displace workers.

——

*$279 billion was the part that went on discretionary spending which put people to work. Much of the rest of the total bill went on tax cuts or Medicaid and unemployment benefits.

**Absurdly, the UK has a shortage of affordable houses, 1.7 million people looking for work, and a shortage of construction skills – all at the same time.

13770284_s

Standard

Part 1 of 2: How Do The Public Understand Economics – And Why It Matters

This is the first of a two-part post. This part looks at why it’s worth studying how the public understand economics, and the second looks at a few things we know about how they do.

Economists have begun to take an academic interest in how non-economists think about the economy.

At first this might sound like a snide dig at economists – a profession seen as hyper-rational turning its intellectual firepower on the persistent question of why non-economists are so irrational. But actually, it’s a very good idea. In a new paper, ‘How laypeople understand economics’, Professor David Leiser of Ben Gurion University and Zeev Krill, a senior researcher at the Israeli ministry of finance, review the literature on the topic, and in doing so compile a good case that what the public understands about the economy has its own economic consequences.

Its message is interesting both to economists and anyone who’s in the business of communicating it.

So – what exactly is the point of studying how the public understands the economy?

Financial literacy

Most prosaically, there is the question of whether we know what we’re doing when we buy financial products. It turns out many of us don’t:

“Recent research has documented great gaps in the ability of savers to manage their savings, due to lack of basic financial knowledge (Lusardi & Mitchell, 2011)”.

Which is an issue, given that:

“recent changes in the labor market, growing availability of debt vehicles like credit cards, and recent pension reforms, in particular the shift from Defined Benefits to Defined Contributions plans, have placed the onus of financial management on the individual consumer.”

This is timely. In Britain, individuals are being given more freedom to spend their pension as they want. This raises the possibility of greater pensioner debt at exactly the time when economists – like Dr Gertjan Vlieghe of the Bank of England’s Monetary Policy Committee – are beginning to recognise that higher debt levels have macroeconomic effects.

Economics Relies on Assumptions about how People React

Many policy recommendations are based on assumptions about how people will respond. But what if their responses are different from economists’ expectations because they understand the economy differently? If economists are recommending policy, they might need to take that into account. 

Take inflation. As the authors put it,

“inflation is perceived as something bad that befalls prices and money: money is worth less, prices are higher. Its consequence is a lower value of the local currency and devaluation.”

Take a scenario where a central bank cuts rates to generate inflation. Now imagine that the public hears of the cut, assumes inflation will follow, worries that their money will be devalued, and concludes they had better save more in nominal terms to stabilise their saving in real terms. If this aggregate saving is large enough to put downwards pressure on consumption and demand, then the cut will produce less inflation than expected. It would take a bigger cut to generate the level of inflation than the Central Bank had expected.

In other words, public perceptions can make monetary policy harder to get right.

Bad for Democracy

An uninformed public can have political consequences.

The paper quotes Philip Inman’s argument that

“when people consistently misunderstand economic debate, it ‘stagnates into discussions between small elite groups over small differences behind the backs of an increasingly disillusioned and unrepresented public. This … is a grave threat to our democracy”

Economists seek solutions, politicians seek votes

I’m sure some people will look down on this kind of research, arguing that academics should be advancing the frontiers of knowledge, not putting their energy towards studying people whose understanding falls well short of them.

But I disagree. Social sciences are not like natural sciences where the pursuit of knowledge is just an end in itself. They also hold out the prospect of being a means to change society through policy.

In democracies, that only works when the policies social scientists recommend can be enacted. But even when social scientists agree on a policy change, politicians will only do it if there are votes in it, and there will only be votes in it if enough people think it needs doing. So to have policy enacted, economists will sometimes find themselves getting stuck into the messy arena of public debate and public understanding.

In short, public misconceptions can severely dampen social scientists’ ability to make themselves useful.

The paper gives the example of pensions. Retirement funding, it points out, is often:

“actuarially untenable, due to a combination of increased life expectancy, lowered return on investments, and demographic changes. In responsibly run countries, this concern requires certain policy changes. The consequence is often that members of the public, who do not understand how pension funds function, feel they are being cheated of their hard-earned rights”.

It also makes the implicit argument that the better the public understand economics, the more likely it is that politicians will. It mentions President Erdogan of Turkey rallying against the governor of the Bank of Turkey for cutting interest rates. Why? Because no matter what econ 101 says, he believed a rate cut would reduce inflation.

(Nor is this example designed to belittle Turkey; Ted Cruz wants to put the US back on the gold standard).

So how do the public understand economics? I’ll look at that in the second post.

Vintage Woodcut Orange

Standard

Part 2 of 2: How Do The Public Understand Economics – And Why It Matters

This is the second of a two-part post looking at a paper, ‘How laypeople understand economics‘ by Professor David Leiser of Ben Gurion University and Zeev Krill. The first part looks at why it’s worth studying how the public understand economics, and this part looks at a few things we know about how they do.

Leiser and Krill show that a good deal of public understanding of economics is based on cognitive shortcuts.

Good begets Good

In a 2014 study, Dräger, Lamla, and Pfajfar used data collected over years by the University of Michigan Survey of Consumers, which asks thousands of people each month about what they expect to happen in the economy.

In defiance of the Phillips curve – the conventional wisdom that unemployment is inversely related to inflation – two thirds of the respondents expected inflation and unemployment to increase together.

This was consistent with a pattern which emerged through the results of other surveys too: people tend to divide economic news into good and bad, and assume that good news causes more good news, and bad news causes more bad.

As explained in the previous post, most people see inflation as bad, because it means their money is worth less. Unemployment is also seen as bad, and so the two are judged to be likely to come at the same time. The authors comment that this system won’t necessarily get you the right result, but it does enable you to answer just about any macroeconomics question.

Metaphor

Much public understanding of economics is based on metaphor. The metaphors we use, of course, determine how we see the problem and the solution. A financial crisis can be seen as a ‘burden, a crime, other people’s suffering, an injustice, an opportunity and a looming threat, an illusion and the doings of fate’. The foreign exchange market can be characterised as ‘a bazaar, as a machine, as gambling, as sports, as war, as a living being and as an ocean.’

And this determines what policy is seen as appropriate.

For example, take manipulation of the foreign exchange market. Should policy try to make the FOREX market fair? If you see it as like sports, you’re likely to say yes. If you see it as like war you’re more likely to say ‘good luck with that’.

Intentionality

But the main way people understand the economy is not as a complex interlocking system of cause and effect, but as the result of the deliberate acts of individuals. The authors call this ‘intentionality.’

They describe a study of public perceptions of the causes of the financial crisis in the USA, Germany, France, Russia, Israel and Sub-Saharan Africa. Across different countries, respondents tended ‘to attribute the responsibility for the crisis to moral, cognitive, and character failures of individuals, rather than to systemic features of the economy’.

People are more likely to see the crisis in terms of moral failures, stupidity, deliberate negligence, lax regulation and supervision than the failure of a system.

And this is surely an important lesson for anyone arguing to change that system: most people, in most places, don’t think there is a system. That is not to say that it can’t be explained. Only that it needs to be.

I’m pleased to see the public understanding of economics beginning to get some scholarly attention, and I’m sure there will be plenty more useful insights from this area in future.

Vintage Woodcut Sliced Orange

Standard