Matt Nolan presents his Top 10 links on the economics of automation

Today's Top 10 is a guest post from Matt Nolan, an economist at Infometrics, and an author at the blog site TVHE.

As always, we welcome your additions in the comment stream below or via email to david.chaston@interest.co.nz.

And if you're interested in contributing the occasional Top 10 yourself, contact gareth.vaughan@interest.co.nz.

See all previous Top 10s here.

Before writing this I had a search of interest.co.nz for “automation”.  I found the following articles as the first 5 links: A top ten by Lena Hesselgrave, an article by Nigel Pinkerton (a colleague of mine), an article from NZIER, and two Top Tens from myself.  Finally, on my travels I also found an article on robots written by me on this site as well! (There's also this from Bernstein analyst Michael Parker, Ed).

So discussing the economic consequences of automation alone is not novel – but given that this is one of the key drivers of changes in the global economy and labour markets, I thought we should give this the full Top Ten treatment.

One thing that is clear is that technological change is disruptive – but it also allows us to make more with less.  As a result, understanding some of the finer points of how this works (in an economic sense) is of use to all of us. That is our goal below.

1. Is automation currently the issue it is made out to be?

Before crystal ball gazing and exploring the futurism of robots, we should root ourselves firmly in the present; is automation showing up clearly in the data now?

For the type of fearful situation of robots replacing workers we keep hearing about, there are two areas we would expect to see robots show up – increased automation for specific jobs and tasks (eg falling employment in a given industry/occupation with a corresponding lift in capital expenditure in this area), and rising overall capital expenditure with an increased pool of unemployed individuals.

The first of these shows us where the shock is taking place. The second shows us that there are no areas where labour has been able to move in the macro-economy.

Although it is clear that the manufacturing industry (and increasingly the service industry) is experiencing automation – it is not clear that automation is really the bug bear limiting employment opportunities that we are often told.  Jared Burstein notes that, in aggregate, the investment figures don’t suggest capital is replacing labour.  Furthermore, as Chris Blatman states there isn’t really any evidence of increased structural unemployment due to these changes.

The most popular work on this issue with reference to automation in manufacturing was by Autor, Dorn, and Hanson (2013).  They found:

“Labor markets susceptible to computerization due to specialization in routine task-intensive activities experience significant occupational polarization within manufacturing and nonmanufacturing but no net employment decline”

In other words, the automation issue is looking like one that changes the tasks and roles people are taking on – rather than an issue that is directly keeping people out of the labour market.  It is with this in mind we can start thinking about the future.

2. Surely the types of innovations matter?

When considering technological innovation economists often focus on what has been observed in the past – automation that is increasingly efficient at performing routine tasks that do not require learning/can be implemented as a rule.  By doing this a role for labour remains. However, Daniel Susskind discusses whether this is really the case – and what the consequences would be if non-routine tasks could be automated, something that is increasingly becoming a reality with technology like driverless cars.

The logic of his piece is that if there is advanced capital that can undertake the same tasks humans can then it is a pure substitute for labour input.  Although such a world involves some positives (output can be created without any work!) this would be the type of situation where a firm would only be willing to use a labour input if the cost was sufficiently low.  If the capital becomes sufficiently cheap, such a world is one effectively without material scarcity, but also without the means for income to be distributed to those who do not own capital.

So if such a situation was to happen what can or should be done?  What about tax?

3. Robots paying tax?

The idea of having people pay tax on capital income is not new – and if the target of a tax system was to raise revenue by taxing income when it is generated then it seems quite natural.

Specifically, Bill Gates appears to be saying there is some level of substitutability between capital and labour – and by taxing the revenue paid to purchase labour inputs but not taxing the revenue paid to purchase capital inputs, we are encouraging more capital intensive forms of production. This would be fine if new job opportunities were being created rapidly enough, but Gates is concerned that they are not.

However, this isn’t quite the full story. Sticking to the same view we can think of the situation as one where the owners of labour (workers) own their labour while the firm owners own the capital – these are called factors of production. Both workers and capitalists then pay tax on the income earned from their factors of production and as a result the robots do pay tax in terms of the incomes they generate for their owners!

We cannot argue it is fair for robots to pay tax as human do – since the income earned from robot work that is not reinvested is indeed taxed. To tax robots, or capital more generally, again implies that there would be horizontal inequity (an issues we talked about here back in 2013).  This doesn’t mean the tax is definitely a bad idea – instead we need a reason why we want to treat robot owners differently to other people when it comes to giving them their tax bill. We will touch on this issue in point 9 below.

Furthermore we need to be a little bit careful when we turn around and tax an input in order to convince people to use another input. When doing this it is because we view them as substitutes – but often capital will work by making labour more productive, what is called labour augmenting change. If robots are performing tasks which are in turn allowing workers to add more value, then taxing to dissuade firms from investing reduces the value of what workers are doing – and in this case the firm will be willing to pay less to hire someone.

In such a circumstance a tax on robots which is put in place to increase wages may well reduce wages – something we need to keep in mind when developing policy.

4. Looking at the past: lessons from the “industrial revolution”.

Having been in the present, gazed into the future, and talked about tax policy, we are now in a place to look at the past.

When working on this Top Ten I was wondering if Tyler Cowen has been talking about this topic recently – it turns out he has had a lot to say.  One especially good piece was when he discussed the current period, which Alan Blinder stated in 2006 would be the “third industrial revolution”, to the actual industrial revolution.

The industrial revolution is so important to look at because of what happened – there was massive increases in economic output, without a corresponding lift in living standards for a large number of people.  From the article:

By the estimates of Gregory Clark, economic historian at the University of California at Davis, English real wages may have fallen about 10 percent from 1770 to 1810, a 40-year period. Clark also estimates that it took 60 to 70 years of transition, after the onset of industrialization, for English workers to see sustained real wage gains at all.

Now this has not happened in New Zealand, but stagnant wages in the United States and the disruptive similarities (in terms of the tasks people perform) between these two “industrial revolutions” should give us pause. Novelists of the day give us a flavour of the abject state of the working classes during this period, with real wage growth only taking off after the Long Depression, a sharp pull back in capital expenditure growth, and the gradual introduction of the welfare state that occurred during this period.

This is not to say that this is what will happen – but it just tells us that we do need to consider winners and losers from automation.

5. The distribution of automation – finding the winners and losers in terms of income/inequality.

Even if we can argue that automation and technological change will drag up wages with it, it is not just the level of income that matters – but also the way it is distributed. 

To think about this issue we can consider what has happened to labour’s share of income/GDP due to technological change.  If the improvement in technology predominately replaced the tasks workers performed, then we would expect labour’s share of GDP to fall. If the improvement in technology augmented current tasks or made it more difficult to produce without labour, then we may expect labours share of GDP to rise. An estimate of this was given by the IMF in 2007.

However, distribution is certainly not about just labour vs capital. The increase in income inequality within countries over recent decades has occurred in the face of growing polarisation in labour markets – a consequence of computerisation as mentioned in point 1. 

We can think about this by considering the process such a change leads to a change in the tasks available.  During the industrial revolution the skilled jobs of many artisans were wiped out by mass production, this led to a polarisation of tasks that unemployed artisans could move into – with lower skilled work on a production line or work that required new skills related to the production line (eg working on maintaining or improving the machine).  This led to a group of middle income jobs being separated out into low and high income work.

Here the artisans lost the value of their skills, and they either had to invest in new skills to receive a higher wage, or accept working in lower income work. Although more efficient production of textiles improved living standards of people who purchased textiles and those who worked in high skill tasks, the people who were pushed from skilled to less skilled work lost out in terms of income – and there is a fear the same thing could be going on in some countries at present.

Furthermore, the type of tasks matter. It can be argued that manufacturing is more egalitarian. On the production line everyone is relatively equal, doing a similar job and receiving comparable pay. With service work there is more competition, more of a push to be a "super star" - in this way I like to view Youtube stars as a dystopian vision of the future of work for us all.

6. The distribution of automation and conservatism – the social consequences of a changing labour market.

Above we talked about the possible economic consequences in terms of incomes and opportunities to earn income. This is a very economist way of looking at things, and misses something that is more important to some people – status and security. 

To see the way that many economists like me communicate – even if we don’t say it too loudly – you can read Betsey Stevenson.

<Policy wonks like me have wondered why more lower-skilled men aren’t adapting. Why don’t they take care of their children when they are out of work? Why don’t they take jobs as home health aides? Or sign up for degrees in nursing? One problem is that these occupations conflict with traditional notions of masculinity. They require sitting, caring and communicating, as opposed to working with big machines.

There is a key point here. The automation of manufacturing work does not mean there are no jobs – and the areas in the US that experienced this have low unemployment rates and have had employment creation. Furthermore, it isn’t strictly about income – with many areas outside of the US that experienced such a change still seeing median incomes rise.

Instead it is about purpose and security. 

Now my problem with how many people like myself talk about this issue is that with our language we will, implicitly, put the blame for the lack of social change on the individual who is disenfranchised. I am not saying that is what we believe (I don’t) – but it is how we sound.

Stating “men need to do women’s work” might be a concise way of stating where the relative opportunities are shifting in areas of the modern labour market – but it shows a certain lack of sensitivity about the way technological change may be undermining peoples sense of worth.

Working in services can involve directly facing clients in a position of relative weakness – because of the way many of us unfairly treat service workers. Add to this a perception about what is appropriate work, a view which is often based on what your parents did, and asking the child of a mill worker to work in a retail firm asks them to sacrifice something about how they saw themselves – and the society they have to make that choice in doesn’t let them forget that.

We are a society of people who, sadly, view our self worth as a function of our status. Yes, people that treat service staff in a dehumanised way are pathetic – but from my experience this isn’t a small part of the community, and it is a cost that people who believed they would be in a different role now have to face.

It is the blindness of analysts and politicians to these feelings that leads to a push towards social conservatism and a desire for things to be how they were in the past – even though technology and increasing rights and opportunities for many groups have been morally good things.

7. Entitlement” and automation.

What happens if automation does take a step up and suddenly there are few tasks for people to undertake – in such a world there would be a growing number of unemployed people unable to find work.

This type of thought experiment makes me consider a term I hear a lot of – entitlement.  Often, with reference to someone receiving job seeker support, I will hear “they feel entitled to that money, but they haven’t done anything to deserve it”.

If that is your view imagine I say the same thing back to you about your own job. No doubt that pisses you off, but give me a second – I wanted to wind you up a bit but now I plan to explain myself!

The access to resources we have from even earning the minimum wage eclipses what we would have had if we put in the same effort 100 years ago. The change in technology that has occurred implies that a lot more can be created, and for the same effort you now receive more.  Why exactly are you entitled to receive some of this increase in output but others who are unable to find work – due to bad luck or sickness – are not?

You may say “because I am working and they are not”.  However, what about if we are in a world where automation has created a dearth of tasks for individuals to undertake and many of those who get remunerated only do so by getting pointless jobs unrelated to these tasks. Excluding those who are arbitrarily excluded from work from social progress appears unduly harsh.

As long as we solely view entitlement to income – which includes the necessities to life – as a product solely of the work you can perform we will continue to victimise those are outside or are on the margins of the labour market.

Personally, I morally agree that someone who sacrifices 40 hours of their week to work should receive more market resources than someone who sacrifices none.  Furthermore, someone who has invested in skills deserves a return on those skills (another area where automation can snag people by destroying their human capital). However, it does not follow that those who refuse (or are unable) to sell their time should be completely excluded from social progress – and I am not sure why so many people from the left to the right feel as if this does follow.

We have a social security system because we believe there is a minimum every citizen deserves as a product of being human – it is a right and they are entitled to it. 

What this minimum entails changes as technology improves and society advances – irrespective of whether the change in technology is changing the availability of work! Between 1989 and 2016 the unemployment payment to an adult individual fell from 38% of average net weekly income to 26%, implying that either our view regarding what is a fair minimum has changed or we have just not realised what is happening.

8. The UBI and automation.

So after we’ve agreed about what minimum people in society have a right to (an open question), how do we ensure people in society receive this minimum?  One solution might be a Universal Basic Income (UBI).

We have had a Top Ten discussion on the Universal Basic Income concept (also called a guaranteed minimum income in New Zealand).  It is not that different from the current benefit system – except the work testing condition is gone, the removal of abatement rates is a significant tax cut for those on lower incomes, and the scheme is relatively expensive implying average tax rates will need to rise to compensate.

Note:  One proviso here is that sometimes when people talk about such a payment they are discussing a fixed sum for everyone. Although this sounds fair at first, it would be much less fair than our current benefit system. Currently payments are partially targeted on the basis of need – for example, someone who is unable to work due to a physical disability also has a higher cost of living and so requires a larger payment to have the same quality of life. Simply giving everyone the same payment would disadvantage the most needy.

With all that out of the way we can ensure that this minimum standard is met by the level we set a minimum income payment – or the level we set current benefits. 

This does help in the case of automation by giving people a minimum they can rely on (security) and removing the social stigma about work in a world where work may be becoming more arbitrary in nature due to the large size of corporations and technological change.

However, it does not solve the issue of dislocation and the lack of purpose – a feeling that the world and technological change has taken away someone’s right to a type of job. This issue obviously matters to many, but the philosophy of life’s purpose is beyond the remit of economists so I will have to leave it here.

9. You would need to pay for such a scheme – how?

Many moons ago I suggested a way we pay for this scheme, which I felt captured the spirit of what the basic income represented. A tax on land. When I suggested having a tax on land it was on the basis that no-one owns land in the same way we own our own bodies, instead the right to the land is given by society in a form of permanent lease. As a result, why not have a payment associated with this lease. 

Although this argument may seem extreme at first, it is not really that novel; Henry George was one of the most vocal proponents of such a tax, but before him Thomas Paine discussed this idea with reference to Agarian Justice, and the rates used to fund local councils work in a similar manner to this.

This could potentially be extended to technological changes. Automation and technology more generally involves some stock of “social knowledge”.  Given this knowledge exists and is in some respects free for firms to use we could similarly argue that they should have to pay some social royalty to use it.

However, as this illustrates we have to be careful. When exactly is an idea fully “social?" What part of knowledge is embedded in the firm and the individuals they hire? Are our non-market choices that use social knowledge liable for tax (eg starting a fire)? Such questions of ownership are difficult, as Tyler Cowen shows when discussing who should own the robots.

10.  But who pays?

Going back to tax, such an argument for taxing capital or for taxing land is an argument for horizontal inequity – we are taxing people more heavily who own land or capital then we are taxing those who generate the same income supplying their labour. We need to discuss whether this is fair, which involves facing the question of whether we believe some part of this land or capital is socially owned.

If technological change is benefiting one group at the cost of another then we can justify taxing one to help the group that is suffering transition. Land and capital taxes will tax all land and capital owners – but does this represent the group we want to redistribute away from?

Currently it appears, at least in the US context, that the beneficiaries of change have been “super managers” and “highly skilled” workers. This does not invalidate a land tax as a potential idea – but it does suggest that we may not be able to justify a tax on the basis of redistribution due to technological change. And before we suggest taxing these people with a more progressive income tax system, we have to ask about the incidence of tax (eg can they pass on the higher tax to other people in the firm, or consumers) and the ability to evade tax.

We also need to accept that, if effective, such a choice of tax and transfers without work testing may necessary lead to lower output – providing income without a jobs test reduces the incentive to work, while taxing capital reduces the incentive to invest.  In a country like New Zealand where land is plentiful a land tax may even lead to lower utilisation of land. 

Although you will often hear people say that such a scheme will “unleash the power of human innovation by taking us away from being a slave to wages”, I suspect people who believe this have an overtly romantic view of what people would do when they are given free access to resources. Status seeking through status goods and an increase in leisure activities are much more likely than a renaissance of innovation – especially in a modern society where innovation is increasingly about group, rather than individual, behaviour.