Andrew Leigh

Race against the machine or race with the machine?

Laying the foundations for an innovative, productive and equitable Australia

Speech - Corrs Chambers Westgarth
Perth

10 March 2015

Nearly five years ago, I made the switch from professor to politician. I’ve never regretted the transition, but it continues to surprise me how different the two occupations are.

In academia, specialisation is valued – you want to pick an area so narrow that you can know more about it than anyone else in the world.

Politics is a generalist occupation – in a typical day, I’d go from talking superannuation to local jobs, from family counselling programs to Free Trade Agreements.

University economics departments value sharp distinctions, and blunt critiques. Politics is a team sport, so you need to build consensus to bring about reform.

One of the really noticeable things about politics is how the language of economics gets used. Terms like human capital, comparative advantage and public goods might be fine for the lecture theatre, but they have a tendency to leave people cold.

When we fail to engage, to explain, and to connect, there is a risk that people go along with economic changes not because they understand them, but because they feel they have no other choice.

Nowhere is this truer than in the case of productivity.

Put simply, productivity is a measure of the amount of output created by a person or a machine over a given period of time.

The typical Australian earns most of their income from work. What we get paid is ultimately tied to how much value we bring to our jobs. So productivity matters.

As Paul Krugman once put it, ‘Productivity isn’t everything, but in the long run it is almost everything. A country’s ability to improve its standard of living over time depends almost entirely on its ability to raise its output per worker.’

Unfortunately, this simple idea – that we can live better lives if we work more effectively – has been lost in debates about productivity.

Too often, people who should know better have tended to make ‘productivity’ synonymous with longer hours and less job security.

If productivity were a marketing concept, its popularity would rank somewhere between frozen berries and giving knighthoods to dukes.

So let’s strip back some of the jargon, and see what productivity means on the ground.

When I visit a new firm, one of my favourite questions is to ask people how they’re doing things better this year than the previous year. So let me give you some examples of where I’ve seen real productivity improve.

Last year, I visited a manufacturing firm that makes mining machines. As you’d expect, machines travelled through the factory on a production line. But as the machines had evolved, the route they followed got more convoluted. The line looped around, doubled back, and covered much of the factory floor. Every extra metre that the line travelled meant more energy to move the machines, and more breaks as workers waited for the next machine to arrive.

So baroque had the production line become that when they revamped the layout, the firm found that they were able to get the same work done on a line just 1/13^th the length. The result was a one-third improvement in productivity for the company.

On another occasion, I visited one of BHP’s major iron ore mines in the Pilbara. In the past, they used to carry out blasting operations every day or two. Blasting isn’t merely dangerous; it is also time-consuming. Everyone except the blast team needs to be out of the pit, and they cannot re-enter until the safety team has ensured that all explosives have been detonated. In recent years, BHP has put considerable effort into increasing the size of their blasts, and making them less frequent. By blasting once a week, they reduce the amount of time wasted.

BHP is also making an effort to improve the time their trucks spend on road; aiming for less than 10 percent downtime. They are working with drivers to help avoid habits that increase wear and tear. As it happens, they’ve noticed that female drivers tend to keep their trucks on the road longer. So they’re looking at breaking that down – figuring out what it is that the female drivers do differently, and how they can teach it to their male drivers.

Another example of how productivity-boosting improvements often come from bottom-up rather than top-down came at Fortescue. They instituted a company-wide competition called ‘Have a Crack’, with the prize being $50,000 for the best productivity-boosting suggestion.

The winning idea was designed to increase the efficiency of the machines that load iron ore onto bulk carriers. As you know, a bulk carrier is made up of a number of different compartments, or holds. When the loader finishes filling one hold, it has to stop the flow of iron ore for a few minutes while it moves to the next hatch.  

In the past, stopping iron ore flowing onto the ship meant stopping it being picked up by the reclaimer. But the winner of ‘Have a Crack’ saw that this didn’t have to happen. He came up with the idea of a surge bin that allowed the reclaimer to store up extra iron ore, ready to go when the loader moved into place over the next hatch.

Fortescue estimated that the idea of a surge bin is worth tens of millions of dollars each year. Not a bad return on a $50,000 investment.

Internationally, economists are increasingly noticing that boosting productivity is about modest tweaks rather than big breakthroughs.

One striking study looked at a factory in the United States that melted down scrap steel into concreate reinforcing bars (rebars) for use in the construction industry. They found that the firm doubled its output in a 12-year period.

[1] As they delved into how the firm achieved such an improvement, the researchers quickly discovered that it wasn’t because people were working harder. Instead, they noted:

‘Conversations with management suggest that they believe the productivity gain can be attributed to “learning through experimentation” or “tweaking the production process.” For instance, experimenting with the way scrap is fed into the furnace and in the timing of the different tasks performed. This form of learning is driven by how production takes place, trying new ways to execute each step of the production process.’

It turns out that this is a frequent story in other industries, including shipbuilding, car manufacturing and textiles. By listening to those on the factory floor, the firm is able to make small, incremental changes that together help raise living standards.

The process of technological change is sometimes posed as a ‘Race Against the Machine’ (to quote the title of Erik Brynjolfsson and Andrew McAfee’s fascinating book). To remind us of how dispensable we are, a recent economics paper estimated that 47 percent of occupations could be replaced by a computer within a few decades. Among the least likely to be replaced were recreational therapists, social workers and audiologists. Among the most likely to be replaced were insurance underwriters, title examiners and telemarketers.[2]

An xkcd cartoon features one character saying ‘Well, at least humans are still better at, uh, coming up with reassuring parables about things humans are better at?’.[3] The comic finishes with a computer running a program to generate thousands of reassuring parables per second.

But there are also instances in which the more appropriate metaphor is a race with the machine. For example, it turns out that computers are surprisingly bad at identifying objects they haven’t seen before. When Google researchers joined 1000 computers together and asked them to identify cats, they got it wrong one time in four.[4] If a two year-old child performed that badly, we would begin to worry about their brain development.

Recognising the strengths and limitations of machines, Amazon has been updating the Kiva system that it uses to pack books. In the past, when a book order came in, a staff member would go off into the warehouse to the location where the computer said it was located. It was hot, tiring work. Now, the Kiva robots go off and get the shelves and bring them to the staff member. The robot does the thing that machines are good at (fetching and filing), while the human does the thing that we’re good at (recognising objects).

One of my favourite examples of productivity growth is a video of Formula One pitstops in 1950 and 2013.[5] In 1950, the world’s best pit crew took a minute to change the tires and refuel a car. In 2013, they took just a few seconds. The difference is due to better skilled workers, better machines, improved teamwork, and information technology. It also turns out that the job of a Formula One mechanic is now safer than in the past.

That change is affecting many parts of our economy. In offices, computerisation means that there are hardly any jobs for typists and filing clerks – positions that were ubiquitous when I worked as a junior lawyer. By contrast, those office support jobs that exist involve more complex interpersonal work, such as planning travel or coordinating meetings.

Will technology keep delivering dividends? At a recent conference I attended at the OECD, this debate was brought into sharp relief by presentations from two of the world’s leading economists.

In the pessimistic corner was Robert Gordon, who pointed out that average productivity growth in advanced countries has fallen significantly over the past decade. Gordon worries that this slowdown is due to structural changes. Other economists have also pointed out that the twentieth century saw an outpouring of innovation, from planes to television, penicillin to microcomputers.[6] On some metrics, they point out, the rate of innovation seems to be tapering off. As PayPal co-founder Peter Thiel puts it, ‘We wanted flying cars, instead we got 140 characters.’

Leading the charge for the optimists was Joel Mokyr, who argues that two systematic factors are fuelling innovation. First, science’s toolkit now includes technologies such as adaptive optics and automatic gene sequencing, which can spur research progress in the same way that the glass telescope underpinned Galileo’s astronomical studies. Second, the internet permits the cross-fertilisation of ideas across disciplines and continents.[7] Mokyr argues that the pace of technological progress is rapid. He points out that people living in innovation ages often don’t realise it at the time. 

As a congenital optimist, I’m naturally drawn to Mokyr’s explanation. But even if technological progress has a bright future, it’s important to consider the impacts on inequality.

In general, economists say that technology is ‘skill biased’. By this, we mean that the beneficiaries of a technological advance tend to be those with the best education.

Since the early-1980s, the number of keyboard operators in Australia has fallen by three-quarters. But we have jobs today that were never envisaged then. On the same block as my electorate office sits 2K, one of Australia’s largest computer game designers, which designed Bio Shock and Borderlands: The Pre-Sequel!. When you walk through their offices, you meet architects, and movement specialists, coders and voice experts. Virtually none of this existed a generation ago.

Over the past generation, we’ve lost over 100,000 jobs for keyboard operators, but created over 200,000 positions for information technology professionals. And on average, the new jobs are more skilled than the old ones.

Innovation also exacerbates inequality in another way. When machines become more productive, their wages don’t rise. Instead, the money goes to the owners of capital. So rapid technological advance not only widens the gap between highly skilled and less skilled workers; it also widens the gap between those who own the new machines, and those who do not.

Like it or not, technological advance and inequality are inextricably bound together. We need innovation as a driver of productivity. But new technologies have a tendency to widen the gap. 

Put another way, many of us are drawn to innovation because we love creativity. But as Schumpeter’s memorable phrase ‘creative destruction’ reminds us, creativity tends to come with destruction.

So what are the answers? Last week’s Intergenerational Report contained some discussion about technological change, but not in enough depth or detail. Building a more innovative Australia is a debate that many of us on the Labor side have been engaged in, including Bill Shorten, Chris Bowen, Sharon Claydon, Tony Burke, Ed Husic, Jim Chalmers, Tim Watts and Clare O’Neil, to name but a few.

If you hear a politician talking about 3D printing, driverless cars, crowdfunding, cryptocurrencies, the sharing economy, electronic health records, nanotechnology or smart grids, it’s more likely than not you’re listening to a Labor MP.

Equally troubling is that the Intergenerational Report failed to properly grapple with the question of inclusive growth – combining productivity with equity. Indeed, the report never once mentions inequality.

This is a pity, because without taking proper account of egalitarianism, it’s too easy to fall into the trap of thinking that the solution to sustainably lifting productivity in lies in cutting wages. Advocates of this approach believe the nation is suffering from a ‘wages explosion’, despite the fact that we’ve just had the slowest rate of real wage growth since 1997. Under this conception, plenty of workers will never produce enough output to justify their wages. Only by cutting minimum wages and eliminating penalty rates – the spruikers say – can we raise productivity.

A better approach isn’t to cut pay, it’s to raise output per person. By boosting the quality and quantity of education, investing in science, improving public infrastructure, and engaging with the world, we can raise people’s productivity. By maintaining a means-tested social safety net, we can ensure that the most vulnerable are looked after. This approach favours a race to the top, not a race to the bottom.

Unfortunately, I don’t see enough evidence that the current federal government shares this vision for the future. Having broken the bipartisan agreement to needs-based school funding, the Intergenerational Report forecasts that education spending as a share of GDP will nearly halve by 2055. With one in five CSIRO staff losing their jobs, we will get less research into energy, forestry and mining. That matters particularly to Western Australia, where science was vital to taking depleted sandy soils and figuring out how to create a grain industry that now produces ten million tonnes annually.

Then there’s investment. As Anthony Albanese reminds me, there’s no evidence that reannouncing Labor-funded roads boosts growth.[8] By building a second-rate National Broadband Network, we limit the opportunities for an entrepreneur who wants to run an online business from their home in Albany. By blocking a foreign investment bid from the United States (for the first time ever), and putting new red tape on housing investment, the government is missing a chance to build our capital-hungry businesses. And you have to wonder – if the government is serious about a targeted social safety net, why do their still have a policy of removing the means-test on the private health insurance rebate?

* * * * *

In this short talk today, I hope I’ve persuaded you that productivity not only matters, but can be interesting. I also hope I’ve debunked the notion of ‘productivity as pavlova’, in which commentators sometimes imply that productivity is a tasty treat that governments simply slice up and hand out to fortunate firms.

Productivity is both tougher and more unexpected than that. It’s the result of process tweaks, unexpected technological breakthroughs, and the application of existing know-how in a new setting. Governments can help create the environment for productivity growth, but rarely do governments directly boost productivity.

At the same time, it isn’t enough to keep looking at the average. We only have to look at the jobs of driving a horse-drawn carriage, carving a sail mast, or hot metal typesetting to know that globalisation and technology are disruptive. With Australian inequality as high as it has been in three-quarters of a century, we cannot ignore the possible impact of innovation on inequality.

In the two hundred millennia since modern humans evolved, there has never been a more exciting time to be alive. There has never been a more challenging time to be crafting social and economic policies to achieve inclusive growth. Let’s hope for the generations to come that we can get it right.