Rodney Dickens notes it has been many years since the warnings of how jobs are under threat from machines first surfaced. He looks at the evidence of what has happened so far

Rodney Dickens notes it has been many years since the warnings of how jobs are under threat from machines first surfaced. He looks at the evidence of what has happened so far

Reports about the number of jobs that will be lost as a result of technological advancements have been doing the rounds for long enough to start testing whether it and the other factors that drive employment are having a net negative impact.

This report tests this by checking whether economic growth has become less employment intensive in NZ, the US, the UK and Australia. It is early days in terms of robotics and AI but the macro-economic evidence points to economic growth being more not less employment-intensive in recent years.

My cynical conclusion is that the technological advancements impacting on especially the growing proportion of service sector workers are having a negative impact on productivity; making economic growth more employment-intensive (e.g. especially younger staff spending more time on Facebook, YouTube, etc. and less working).

However, this report is not designed to provide a comprehensive insight. It provides early macro-economic evidence that brings into question the threat to employment from technology.

Warnings about the threat to jobs from the likes of robotics and AI are not new

On a reasonably regular basis a new report about the threats to jobs from technology like robotics and AI surfaces. The following is a relevant quote:

"Brynjolfsson, a professor at the MIT Sloan School of Management, and his collaborator and coauthor Andrew McAfee have been arguing for the last year and a half that impressive advances in computer technology—from improved industrial robotics to automated translation services—are largely behind the sluggish employment growth of the last 10 to 15 years. Even more ominous for workers, the MIT academics foresee dismal prospects for many types of jobs as these powerful new technologies are increasingly adopted not only in manufacturing, clerical, and retail work but in professions such as law, financial services, education, and medicine."

Similarly, a study by Oxford University academics concluded that:

"About 47% of jobs in the US in 2010 and 35% in the UK were at 'high risk' of being automated over the following 20 years."

These reports could have been written at any time in the last few years but were written in 2013 (i.e. probably long enough ago to start testing whether the dire predictions were accurate).

Evidence from NZ suggests GDP growth has become more employment intensive

If technology poses a growing threat to jobs there should be a growing gap between economic/GDP growth and employment growth (i.e. progressively fewer employees should be needed to underwrite growth in economic activity).

To test this, the top right chart compares NZ GDP and employment growth. The best fit, with a high correlation of 0.91, is with GDP growth leading by one quarter; reflected in the green GDP growth line being advanced or shifted to the right by one quarter. On average since 1992 GDP growth has been 1% (i.e. one percentage point) stronger than employment growth due to productivity growth that includes the impact of technological advancements.

If technological advances posed a growing threat to jobs, economic growth should become less reliant on employment growth. The gap between the two should grow. Instead it has narrowed in recent years (second chart below - the green line shows the average gap since 1992).

If technology was starting to have a major impact to the extent it overwhelmed other factors at work the gap should have become more not less negative. At the macroeconomic level the evidence points to economic growth becoming more labour-intensive not less intensive since MIT and Oxford University academics issued the dire warnings in 2013.

Other factors are at work including employment by industry evolving partly as a result of technological as shown in the next chart.

Manufacturing and primary industry jobs have been losers as a result of globalisation and technological advancements. Service sector jobs have been the main winner while government-related jobs have been relatively stable as a percentage of total jobs.

My tentative hypothesis is that the evolving make up of employment towards service sector jobs has contributed to economic growth becoming more job intensive; more than offsetting any negative impact of technology. A cynical view is that technology is having a negative impact on productivity especially in the service sector because a growing number of workers are Facebooking, Instograming, Tweeting, YouTubing and Amazoning instead of working.

It is still early days in the adoption of robotics and AI while maybe NZ is a laggard on these fronts. So it isn't possible to conclude with confidence that all the factors at work including technology don't pose a net threat to employment. However, the early macroeconomic evidence points to GDP growth becoming more not less jobs-intensive.

The US, UK and Australian experiences are roughly similar to the NZ experience

The next two charts show that US GDP growth has become more employment intensive since the dire warnings were issued in 2013; the gap is narrower.

The next two charts show a vaguely similar story for Australia. The peak correlation at only 0.63 is with GDP growth leading employment growth by three quarters. The resource sector has impacted in Australia; it is a sector where jobs have been stripped. However, despite this, most recently GDP growth has exceeded employment growth by less than average, consistent with NZ and the US.

The next two charts show a similar story for the UK with GDP growth leading employment growth by two quarters and the gap being smaller in recent years not larger; roughly consistent with the NZ, US and Australian experiences.

Often too much is attributed to the topical factor when other factors matter more

Technological advancements have and will have a major impact on jobs. This is clearly evident in NZ with it partly contributing to the shrinking share of manufacturing and primary sector jobs. However, the evidence so far from NZ, the US, the UK and Australia suggests GDP growth has, if anything, become more not than less employment intensive over recent years. This is contrary to the warnings about technology-driven job losses that are issued from time-to-time by academics and others.

Like many things, what drives employment is multi factorial and even economists who supposedly study such matters seldom have a real grasp of the factors at work and the impact they will have over the next year let alone the decades ahead. As has been the case in the past, technology and other factors will drive changes in the composition of employment. Technology could mean the speed of such changes will increase. For many individuals technology will pose a threat to their current jobs, but this doesn't mean it necessarily poses a threat to employment in general.

Things don't change fast enough on this front to warrant it being revisited regularly but my plan is to revisit the macro analysis in this Raving every few years to see if technological advances and the other factors at work are contributing to economic growth becoming more/less employment-intensive.

Over the next few years changes in labour market laws in place or planned in NZ will encourage some employers to substitute technology/robots for staff. But I am a bit of a sceptic regarding how much this will impact on employment in part because of the other factors that will be at work. Based on the experience in recent years the other factors have made economic activity more labour-intensive.

This article is re-posted here with permission. The original is here.

We welcome your help to improve our coverage of this issue. Any examples or experiences to relate? Any links to other news, data or research to shed more light on this? Any insight or views on what might happen next or what should happen next? Any errors to correct?

We welcome your comments below. If you are not already registered, please register to comment.

Remember we welcome robust, respectful and insightful debate. We don't welcome abusive or defamatory comments and will de-register those repeatedly making such comments. Our current comment policy is here.


Comment Filter

Highlight new comments in the last hr(s).

Apart from self-checkouts and ordering kiosks at McDonald's I haven't seen much visible technology that directly replaces employees.

Chat bots leave me distinctly unimpressed.

Probably the first technology that could have a substantial widespread and swift impact on employment will be self-driving cars. Uber and others will be quick to roll out fleets of these as taxis, and long-range trucking is already starting to be automated in the US for this (automated trucks drive over interstates at night between depots, where humans take over for the last 20 miles of city driving). Couriers have a bit more job security since the driver also gets out of the van to deliver and collect parcels, although it's not infeasible for this to become more self-service where a parcel recipient walks out to the unmanned van to collect their parcel, but obviously this has several more issues that need to be overcome. Delivery via drone could potentially leapfrog those issues but that seems like 5+ years away for and meaningful commercial adoption.

Another technology that is having an impact, although likely too small to be measured in NZ just yet, is that electric vehicles need quite a bit less maintenance than petrol ones.

Soon you won't be able to tell what is a chat bot and or whether you're talking to a real human:

Interesting article. Yes there are many factors at play but the transition from manufacturing to a service based economy would be a big culprit in my eyes as it’s not anywhere near as scalable thus reducing the employment to gdp gap.
Ironically, maybe AI and automation actually has the potential to bring manufacture back to NZ as it wouldn’t be using many if any of our entry level yet high cost workers. We will have to wait and see I guess.

I don't think NZ has the talent, infrastructure, industrial might or access to capital it takes for companies to set up manufacturing operations here.
Due to abject failure of successive governments and short-termist business groups in keeping us ahead of the innovation curve and preparing ourselves for Industrial Revolution 4.0, I think we are very likely to sit this one out.

Yeah, nah, it is already happening.

Service industry growth? Perhaps. But in my view it is tied up with increasing narrow ownership of enterprises. We have allowed the re-emergence of oligarhcy.
The big corps, are interested in control. The don't make money from innovation and productivity. They also work best with controllable low skill and interchangable workers. (ever tried proposing an innovative solution to the voice at a phone company call center) (Chorus and it's fibre install. Low cost now, substandard work, excessive life time cost, worker exploitive)
Get rid of the wasteful monopolies, including big government, allow small business ownership to blossom, and folk will be employed and reap the benefits.

Getting rid of monopolies in the tech age won't be an easy task, in fact, we could witness more corporate consolidation in the not-so-distant future. The robotic stakes are stacked against small businesses because of the high adoption cost of AI for first-movers.
Moreover, large corporations have access to seemingly endless piles of cash at dirt-cheap rates owing to years of easing.

I love the way that some Economists and Tech Platform owners try to justify the hollowing out of the Middle Classes careers with the "But there will be jobs in the future that we haven't even thought of yet" Or "People will have multiple careers through their life times" arguments. Have they even considered the time and the cost of retraining in to other skill set areas, and whether that will be even feasible for most people?

Because it's not just the low paying manufacturing jobs that are being effected, or even basic administrative jobs. It's high level skill careers that are in the process of being restructured, such as; Doctors, Lawyers, Teachers, Bankers etc.. Basically careers that involve a high level of knowledge retrieval can be replaced by AI.

But that's ok, the big corporate tax dodging tech companies will tell you that everything will be fine....

It's not early days for AI. I saw a japanese busbar controlled by AI logic 30 yrs ago. Talking about the electricity industry, the wholesale market brought in by the accountants has vastly increased the number of analysis in the industry and has resulted in duplication of effort where every generation company needs it own portfolio of potential future projects. The problem is that they are mutually exclusive!

Fuzzy logic != AI. 30 years ago there wasn't the available computing power for anything worth calling AI.

The most important technical advancement is not actually the computing power itself, but the datasets, both their availability and the ability to simply store the volume of data required for training of neural networks.


Can I ask what made you pick this? Are you involved with the rare earth elements professionally?

AI is hard to define. If you think of it as analagous to lever making a man stronger but making his brain smarter then the first program that could be considered AI was a checkers playing program back in the sixties or mid fifties. For more recent accomplishment AlphaZero is giving chess players something to think about.
AI has been able to diagnose some illnesses better than expert doctors for a long time but (a) it has had a narrow focus and (b) it has been unable to explain how it arrived at the diagnosis is a manner that normal humans can understand.

I bet a million dollars you would see a different picture if you studied Japan, they are highly reliant on automation for gdp growth as they have a stable population and an advanced technological society.

Excellent article.
I've seen plenty of evidence that millenials, in particular, are on social networks quite a lot during working hours....and that's in an office setting, not service sector

I’m guilty of that!

"My cynical conclusion is that the technological advancements impacting on especially the growing proportion of service sector workers are having a negative impact on productivity; making economic growth more employment-intensive (e.g. especially younger staff spending more time on Facebook, YouTube, etc. and less working)."

A very cynical view indeed. I would love to see the author spend a month working in the average call-center. He won't have time to go the bathroom, let alone waste time on social media. I bet none of his co-workers back in his days ever went on smoke breaks or had chats with their coworkers.
My cynical view is that the main difference between "wasting time" at work today vs a decade or two ago is not the amount of time spent not working, rather what people spend that time on.
I think the chart about employment by industry groups shows the real reason for no significant drop in employment: machines have replaced workers in manufacturing, but aren't smart enough yet to replace people in other sectors.

The rate of change in capability of deep learning is hard to overstate. At a recent AI research conference median prediction for human level AI was by 2030!! That should absolutely scare the hell out of everyone because 1-5 years after that they are likely to be a million times smarter than us (due to accelerating self-improvement as general AI rapidly begets further improved general AI).
- Neural net processing power has increased about ~1million times in last 7 years, and is anticipated to do another ~100000x in next 4 years.
- It is now possible to run neural nets of similar complexity and processing power to a human brain (~100pflops) affordably eg Google's TPU 3.0 'pods' can do about 90 pflops. Running a human-childhood sized training program is still unaffordable (biggest training effort to date was about 20 days worth of human brain-life)
- Neural nets can now transcribe speech and translate as well as humans, object recognition is as good as humans (depending on training data).
- Neural nets are displaying ability to spontaneously create classifications/groups for new things they are exposed to - in same way our brains do.
- 2 weeks back a neural net won against best professional players in world at Dota2 (warcraft like real time strategy game, with restricted view of world, need to guess at opponent's behaviour, long time horizons for payoff from any particular action, and up to 10000 'moves' per game.

Until about 5-6 years AI was almost nowhere. With complexity similar to insects, now they are about same as small rodents, and in a few years will be similar to dogs and then a few more years humans. There are still problems to solve with regard to structuring them, and figuring out how to coordinate lower level functionality with higher level general intelligence type executive control, but there are lots of ideas on how to do that that haven't even been attempted yet as have only just hit necessary levels of complexity. There are a tonne of hints of what to do from neuroscience and 1000's of researchers gunning to solve the problem with 10's of billion in commercial and govt R&D funding (playing for all the marbles at this stage), so it is likely just a (short) matter of time. General AI is coming, almost certainly within your lifetime, and in many cases likely before your retirement. We cannot stop it as it has such tremendous commercial and military advantage (first superpower/multinational with AI can likely dominate others, so they are all racing for the cliff)

In the meantime the rate at which they can supplant human human workers is nearing an asymptotic line. In 3-5 years they will be controlling robot arms with human like dexterity: and with modest task-specific trained intelligence.
Mobile dextrous robot platform (spot mini):
That is sufficient to replace (eg) alost all hands-on horticultural workers, checkout operators, cleaners, laundry services, much food preparation, warehouse operation, grounds keeping, butchery, packing, goods delivery, repetitive manual manufacturing processes, much building work. Not to mention autonomous driving (Tesla claiming level 5 autonomy within 2 years with their new 144Tflop (0.5% of human brain, similar to a rat or an owl) in-car computer, I suspect autonomy will need another order of magnitude increase to bring it up to 'cat' sized, but that is only 2-3 years away)

The hardware is basically 'there' now. It will just take a couple of years for the development and roll-out into the world. All previous industrial revolutions were about replacing human muscle with machinery. Now we are replacing the human brain. There is just no place left for all the humans to go to make the new jobs required, and where previously only the few percent with IQ<75 were effectively unemployable, soon that number will start to rise, rapidly. There will be no jobs for most kids now in school.

All other problems in the world are as nothing compared to this, and all other problems with a time-line of decades (demographic, economic, environmental, political) are pretty much pointless to be concerned about. All the rules are about to change in an impossible to predict manner. I implore you to start mustering an appropriate rational emotional response and taking it really seriously as the supreme existential threat it is:

I agree, AI is getting very scary. Humankind hasn't experienced anything comparable to it, and in my opinion it's analogous to meeting a highly advanced alien species. The good thing is, at the moment AI is bound to a hardware that a human can turn off with the flip of a switch.
Max Tegmark's book "Life 3.0" describes a lot of possible outcomes, I highly recommend reading it if you haven't already done so.

The AI box problem: The AI can offer you immortality, extreme riches, or threaten you and family with eternal torture in simulated hell (either by itself or AI's that follow it), so not easy to turn it off. You are an ant playing Gary Kasparov at chess, and the AI, with no limit to how long it lives can be very very patient in achieving it's ends.

Very interesting debate, well worth a watch: Be afraid, be very afraid: the robots are coming and they will destroy our livelihoods