ROBOTS, REMUNERATION AND RESTRUCTURING: HOW DO TECHNOLOGY AND INEQUALITY SHAPE ONE ANOTHER, AND WHAT SHOULD WE DO ABOUT IT?*
Annual Sir Leslie Melville Lecture
Australian National University
Sir Leslie Galfreid Melville was a remarkable Australian. Born the year after federation. Trained in engineering and science before wisely settling on economics. Inaugural professor of economics at Adelaide University at age 27. Founder of what would become the Reserve Bank’s research department. Leader of Australia’s delegation to Bretton Woods. ANU Vice-Chancellor for most of the 1950s. Appointed to chair the Tariff Board by McEwen, it is to Melville’s enduring credit that he quit the post rather than succumb to McEwenism.
Having been born at the dawn of the twentieth century, Leslie Melville lived to see the start of the twenty-first. As one obituary noted, ‘there has not been another Australian economist to hold the range of jobs that Melville did’.
It is virtually impossible to think about Melville’s life without being conscious of the technological changes that took place during it. The twentieth century – or the ‘Melville Era’, as Australian economists might call it – saw an explosion in technologies. In transport: planes, helicopters, mass-market cars and space shuttles. In communications, radio, television and the Internet. In health, antibiotics, sewered cities, the pill, and genetic engineering. Not to mention atomic bombs, vacuum cleaners, smartphones, radar, the bra, and plastic. And yet for most of the twentieth century, we not only saw rising living standards, but falling inequality. Melville’s working years – the 1920s to the 1970s – saw the largest reduction in inequality in Australian history.
My focus today is on two challenges: how do we continue the pace of innovation in the twenty-first century that we saw in the twentieth? And how do we ensure that prosperity is broadly shared? By acknowledging the tendency of technological change to increase inequality, we can harness the gifts of Prometheus without suffering their destructive tendencies. As it happens, I will argue that a single policy recommendation offers the greatest promise to make us more entrepreneurial and more equal.
Let’s start with innovation and entrepreneurship. Innovation is the production of new ideas, while entrepreneurship is the process of turning those ideas into products or services that can underpin a successful business. It’s commonplace to say that Australia produces our fair share of ideas, but underperforms when it comes to turning ideas into start-ups. But as I’ll argue, we need more innovation and more entrepreneurship.
Now, I must admit that I’m not the first parliamentarian to discuss this issue. Across Australia’s start-up incubators, visiting politicians have become as common as cans of Diet Coke and Red Bull. This is no bad thing – new business formation is as critical to economic growth today as it ever was. But if we’re devising policies to boost innovation, they need to start with a clear-eyed analysis of the data.
So let’s run through a few numbers.
When it comes to the most-cited research in science and engineering, Australia accounts for 5.5 percent of the most-cited papers (around ten times what our share of world population would lead you to expect). Across science fields, Australia does especially well in geology, geochemistry, earth sciences, environmental sciences and veterinary sciences. Australia has more top researchers per person than almost all advanced countries.
However, we do less well when it comes to translating those ideas into businesses. According to former Chief Scientist Ian Chubb, just 1.5 per cent of Australian companies developed new-to-the-world innovations, compared to between 10 to 40 per cent in many other OECD countries. Just 6 percent of ASX300 firms say that Australia is a ‘highly innovative’ nation.
A similar picture emerges when we look at the workforce. Researchers make up 0.85 percent of the Australian workforce – about average for advanced nations. But the share of those researchers in business (32 percent) ranks us the lowest in the OECD. We also have one of the lowest rates of industry–research collaboration in the OECD.
Another indicator is to look at patent filings. Whether or not you think our intellectual property system is perfect, the number of patent filings does say something about the underlying level of innovation in the country. According to the OECD, Australia ranks in the bottom half of advanced countries for patent filings per person. Over the past decade, most advanced countries have increased their rate of patent filings. But the number of patents filed by Australians is down by at least one-fifth over the last decade.
So a comparison with the rest of the advanced world suggests that we do better on idea-generation than business development.
However, it doesn’t follow from this that policymakers need only focus on the start-up side. Think of it this way. Suppose Australia is producing 100,000 clever ideas each year, and 10 percent are becoming new businesses. If we want more business ideas, we can either produce more than 100,000 ideas, or raise the commercialisation rate beyond 10 percent. Most likely, we’ll want to do both. As I will argue below, this is fundamentally an issue of human capital – ensuring the Australian workforce has the necessary skills to forge new ideas and turn them into great businesses.
Over recent months, Bill Shorten’s Labor Opposition has proposed some ideas to boost innovation and entrepreneurship. These include promoting the teaching of coding and computational thinking in schools, offering a startup year to young university graduates seeking to start their own enterprise, creating new visa categories to attract entrepreneurial immigrants, and getting startups to help solve government problems through Challenge Platforms.
Over the longer term, we also need to consider whether we need to update other parts of our regulatory framework. Does it really make sense to allow firms to include clauses in employment contracts that prevent employees from leaving to start a competing business? Is a patent system that gave Amazon a 20-year monopoly over ‘one-click ordering’ appropriate to the modern economy? Should we be worried when Google tells us that Australia’s copyright system would not have allowed them to create their search engine here?
In other areas, an increasingly weightless economy is creating measurement challenges for government. Australians such as Colin Clark led the way in the development of national income accounting. But increasingly people are asking whether Net National Disposable Income Per Capita would be a better measure of wellbeing than Gross Domestic Product. And in either case, the challenge remains of ensuring that a statistical framework developed for the wheat and steel era can continue to give us a good picture of economic activity in the world of the sharing economy.
From innovation to inequality
But while innovation and entrepreneurship are important to increasing total output, it does not follow that this prosperity will be broadly shared. I recently took a taxi to a Sydney event on the sharing economy. On the way, my taxi driver talked about his concerns on how ridesharing would affect his work. At the event, most entrepreneurs were positive about the quality of services available in the sharing economy.
One innovator’s disruption is another person’s job loss. The Luddites may have used a little too much force to make their point, but they weren’t wrong about the fact that technological change can destroy jobs as well as create them. So it’s useful to consider what changes technology has wrought across the job distribution.
In a series of articles with not-so-reassuring titles like ‘Why Are There Still So Many Jobs?’, David Autor and co-authors have divided jobs into three categories: manual, routine and abstract. They show that routine jobs, which tend to fall in the middle of the wage distribution, have shrunk most. This is true in the US for the 1980s, 1990s and 2000s. It is also true of most European countries over the period 1993 to 2010.
Routine jobs are occupations such as bookkeeping, administrative support, and repetitive manufacturing tasks. What makes routine jobs vulnerable to computerisation is that they involve following established rules.
By contrast, abstract jobs involve problem-solving, creativity, and teamwork. When a lawyer advises a client whether to accept a plea bargain or a manager decides how to respond to an employee arriving late for work, they are tackling problems that do not have a closed-form solution.
More interesting is the resilience of manual jobs to computerisation. Thus far, attempts to automate the work of jobs such as cooking, cleaning, security work and personal care have largely failed.
What does the picture look like for Australia? Fortunately, two recent studies have taken a careful look at precisely this question. Analysing broad occupational data from 1966 to 2011, Jeff Borland and Mick Coelli analyse occupational growth across ten job categories. Ordering the occupations from lowest hourly wage to highest hourly wage, they are then able to look at job growth in the best-paid and worst-paid occupations.
They find that the share of people working two best-paid occupations (managers and professionals) rose by 14 percentage points, while the share working in the two lowest-paid occupations (sales and food/cleaning) rose by 3 percentage points. Yet the two occupational categories in the middle of the distribution (production and operators) shrank by 20 percentage points.
An alternative approach is to look at a shorter timeframe, but more fine-grained job categories. Analysing employment changes from 1993 to 2013, Roger Wilkins and Mark Wooden look at employment changes across 43 occupations. This shows the same broad U-shaped pattern, with growth in both low-paid occupations (such as storepersons and carers), and high-paid occupations (such as managers, ICT professionals and legal professionals). In the middle, jobs like clerical workers, machine operators and secretaries have shrunk relative to the workforce as a whole. But there are also some exceptions to the story, such as farm workers (low-paid, but shrinking) and sports workers (mid-paid, but growing).
Finally, Jeff Borland and Mick Coelli seek to directly address the question by coding jobs according to their degree of routine content. They find a strong pattern: the more routine a job is, the more likely it was to have shrunk since the mid-1960s.
This process can be seen continuing in a host of different labour markets. For example:
- A fast-growing sector of financial advice is the phenomenon of ‘robo advisers’ – computer algorithms that seek to compile and manage an investment portfolio to suit individuals’ tax status and state of life.,
- On farms, unmanned helicopters can now be programmed to spray pastures, at a fraction of the cost of piloted crop-dusting aircraft.
- Companies are continuing to automate their human resources and back office functions, removing mundane and routine tasks such as payroll and invoicing.
- 3D printers are now being adapted for use by the construction industry, to ‘print’ large structural parts. The ultimate goal is to print an entire building on site.
What should we do about it?
At this point, I had been tempted to discuss half a dozen policy ideas that might either make us more innovative, or more egalitarian. But instead, I want to take a different approach. I want to propose a single idea that I believe would radically transform both our innovation landscape and our level of inequality.
If you speak with start-up founders, and ask them what would make their business more productive, they invariably point to the need for highly skilled people. If you chat with people who do job placement, they’ll tell you about the value of a great education as a platform for lifelong learning. If you sit down with those who’ve spent time behind bars, they’ll often talk about how much they hated school.
Over recent decades, we’ve done a reasonable job of increasing the quantity of education received by the average Australian. Since the mid-1970s, school completion rates have risen from three in ten to eight in ten, while university attendance rates have gone from one in ten to four in ten.
But we haven’t done as good a job at education quality. Melbourne University’s Chris Ryan and I showed that if you looked at the literacy and numeracy of 14-15 year olds, it had either flatlined or fallen slightly from the 1960s to the early-2000s. Over the past decade or so, the international PISA exams show Australian 15 year-olds going backwards in maths, reading and science.
For example, over the period 2003-2012, Australia’s average PISA score in maths slipped from 524 to 504. Michael Thawley, head of the Department of Prime Minister and Cabinet, recently described this as ‘a 300-500 percent worse result’, which didn’t quite nail the magnitude, but did inadvertently remind us that the nation’s numeracy problem goes right to the top of the public service.
Most recently, NAPLAN figures for grades 3, 5, 7 and 9 show little evidence of improvement over the period 2008 to 2015. To the extent that there has been improvement, it tends to be at the lower grades, suggesting that preparation for school has improved, rather than value-added at school.
The most significant reform during this period – the move towards needs-based school funding – only began with the 2014-15 financial year, so it would be unrealistic to expect it to show up in student results immediately. I strongly support needs-based funding, and believe it would be a mistake to scrap it. But I also think getting the funding model right is only part of the answer to building a great education system.
To raise the quality of Australian education, I believe we need a national push to raise teacher effectiveness. As my colleague Kate Ellis puts it, ‘teachers don’t just help students build skills, they change lives’. In the past, Australian education ministers have noted that ‘teacher quality’ is the single greatest in-school influence on student engagement and achievement. In my view, this means it should be a central priority for Australian schools policy.
First, some background. In the 1950s, Australian teacher aptitude was underpinned by one factor. Rampant gender pay discrimination in law, medicine, business and other professions meant that teaching was one of the few professions open to female university graduates. Teaching wasn’t free of discrimination, but it provided considerably better opportunities for talented women than other occupations.
Gender pay discrimination meant that the calibre of many professional occupations was lower in the 1950s and 1960s than it would otherwise have been. For example, as late as 1971, women made up just 13 percent of doctors, 6 percent of lawyers, and 0 percent of members of the House of Representatives.
The reduction in gender pay discrimination was one of the great post-war advances for Australia. Not only did it expand the opportunities available to women, but it raised output, because discriminatory firms are less productive. Of course, gender pay discrimination still exists, but our public services and businesses are more productive now than in the Mad Men era: because they make better use of the talents of women.
Yet as talented women flooded into the professions, they flooded out of occupations such as teaching and nursing, which had been relatively attractive to women. Melbourne University’s Chris Ryan and I look at the literacy and numeracy of new teachers, relative to those in their age cohort. We estimated that from 1983 to 2003, the average percentile rank of those entering teacher education fell from 74 to 61, while the average rank of new teachers fell from 70 to 62. The share of new teachers who were in the top fifth of their class halved. The share of new teachers who were in the bottom of their class doubled. (Note that these are all comparisons relative to the cohort, so they are not driven by the across-the-board drop in standards that I outlined earlier.)
The Leigh-Ryan study only covered the period up until 2003, but there is little evidence that teacher aptitude has risen since then. Over the period 2005 to 2012, the share of teacher education students with ATAR scores over 80 fell from four in ten to three in ten.
Having noted these figures, it’s important to recognise that academic aptitude isn’t everything. After all, a good player doesn’t always make a good coach. My favourite teachers were inspirational because they loved their subject matter. To hear my English teacher, Judith Anderson, talk about Donne was to be transported back to 17th century England. Once my maths teacher, Mick Canty, taught you why complex numbers existed, you felt you’d known it all your life. Passion, empathy and grit are all elements of terrific teaching. But aptitude matters too; great teachers are more commonly those who aced a subject rather than barely passing.
It’s also important to note that a statement about averages is just that. To note that the average academic aptitude of new teachers is falling is not to suggest that we don’t have great teachers in the system today. People like Geoff McNamara, a former optical mechanic who teaches at Melrose High in Canberra. McNamara has published over 100 articles and three books, on topics such as dark matter, pulsars, and gravitational waves. He not only won last year’s Prime Minister’s prize for excellence in secondary science teaching, but has inspired an ‘I was taught by Mr Mac’ Facebook group. Geoff McNamara is a sensational teacher, and we need more like him.
In considering how to raise teacher effectiveness, we could do worse than to look to the performance of Finland, a country that routinely ranks near the top of the international test score league tables.
Now I admit that as an education researcher, I’d always been a bit sceptical of those whose education reform ideas seemed to boil down to ‘be like Finland’. Somehow, they always reminded me of Monty Python’s song ‘Finland, Finland, Finland – Finland has it all’. The problem is that it isn’t clear which bit of Finland’s approach we should replicate: Late school starting ages? Less homework? Long recess breaks? No school uniforms? Low levels of inequality? A logical language, in which words are pronounced as they are written?
But what’s far more interesting to me than ‘Finland is great’ is the fact that this is only really the story of the past generation. In the 1960s and 1970s, Finland was an average performer on international tests. Indeed, one study estimates that Australian students outperformed their Finnish counterparts in the mid-1970s. It was only in the late-1970s that Finland embarked on a major push to raise the aptitude of new teachers.
One mark of the success of Finland’s teaching push is that teacher education students are generally drawn from the top fifth of high school graduates. For every position in a teacher education course, there are around ten applicants. Those selected are chosen based not only on academic excellence, but also on an interview. Finnish teachers are highly regarded, with polls placing them as the nation’s most admired profession.
Raising teacher effectiveness involved a suite of changes in Finland. Smaller teacher education providers were closed down. The remaining universities were forced to be more selective and rigorous. The government worked closely with the teacher union in implementing the changes. Teacher pay is about average for the OECD, though the wage gains to experience are relatively large. Finland has no national system of teacher merit pay, though municipalities sometimes pay bonuses to high-performing teachers.
Finland also mandated Masters degrees for all teachers, meaning that those who wanted to teach needed to study for a minimum of five years. Economists have generally been sceptical of the value of Masters degrees, noting that long study periods can act as a barriers to entry, effectively discouraging talented people from entering a profession. Studies in the United States and Australia have failed to find evidence that teachers with a Masters degree do better in the classroom.
Why was Finland’s push for Masters-trained teachers a success? One possibility is that Finnish Masters degrees were more focused on improving teaching than those in other countries. For example, students studying to be a secondary school teacher spend one-third of their time during a Masters degree teaching in schools. Another possibility is that Finland succeeded despite its emphasis on Masters degrees. Once you select teacher education students from among your best high school graduates, it is plausible that what you do with them at university is of secondary importance.
Indeed, this goes to a broader point. Once you select superstar teachers, many of the critical issues in education become less important. I’m a strong supporter of a national curriculum, test score reporting and raising the calibre of school leadership. But I have to admit that all these issues become less critical the more effective our teachers are.
How might we transform teacher effectiveness in Australia? The first answer is to make it our foremost schooling priority. For all the talk about teaching being the number one factor in school effectiveness, teacher effectiveness risks getting lost in the weeds of a host of other agendas.
Making teacher effectiveness the leading priority means doing more to celebrate great teachers, and recognising their power to transform lives. In 2006, ANU invited graduating students to nominate a school teacher who had made a difference. To read the nominating statements is to be reminded of what great teaching involves. One female student wrote of her biology teacher, Lorraine Huxley, that she had ‘inspired me to include science as part of my tertiary studies’. A commerce student wrote of his teacher, David Dorrian: ‘[he] was truly excited about his topic – Mathematics – and imbued this in his students’. An archaeology student wrote of her ancient history teacher, Mary Condon: ‘Mrs Condon had a way of making historical characters and events come alive. I felt like I knew exactly what it was like to grow up in a Spartan village and I felt that Augustus and Agrippina were close friends of mine.’
Imagine how much more innovative and egalitarian our schools could be if every student felt like this about every teacher.
In his 1993 interview with Selwyn Cornish, Leslie Melville reflected back on his many decades of practicing economics in Australia – including interacting with Keynes, leading the Australian team to Bretton Woods, and becoming ANU’s second vice-chancellor. Asked about the role of economists, Melville replied ‘Certainly economists are having a much greater influence today than they did 60 to 70 years ago. They have, for example, miraculously persuaded people that tariffs ought to be reduced.’
As an ‘elected economist’, I occasionally share Melville’s sense of wonder – the miracle that a profession that communicates in fluent Greek and broken English ever manages to persuade policymakers of anything. And yet the reality is that economists really have managed to make a mark on modern-day Australia. From tariff cuts to income-contingent loans, central bank independence to universal superannuation, our nation is happier, healthier and better educated as a result of ideas formulated by economists.
As we grapple with the twin challenges of boosting innovation and reducing inequality, Australia will again need to draw on economic expertise. A major increase in the quality of school education would produce outcomes that would dwarf minor tweaks to our tax or regulatory structures. As the late James Tobin once noted, ‘It takes a heap of Harberger triangles to fill an Okun gap.’
Education remains the best productivity-boosting policy – and the best antipoverty vaccine – that we have yet developed. To transform teacher effectiveness would both raise the rate of innovation and entrepreneurship, and reduce joblessness and inequality. If the robots are coming, we’d best ensure we have great teachers to meet them.
* Thanks to Jason Clare, Kate Ellis, Daniel Prager, Jennifer Rayner and Tim Watts for valuable feedback on an earlier draft, and to Mark Wooden for sharing data on occupational changes. Amy Haywood, Joseph Walker and the Parliamentary Library provided outstanding research assistance.
 John Farquharson, 'Melville, Sir Leslie Galfreid (1902–2002)', Obituaries Australia, National Centre of Biography, Australian National University. Available at http://oa.anu.edu.au/obituary/melville-sir-leslie-galfreid-720/text721.
 Office of the Chief Scientist 2014, Benchmarking Australian Science, Technology, Engineering and Mathematics. Australian Government, Canberra, p.v.
 NAB Group Economics, ‘NAB Special Report: A Survey of Business Innovation in Australia: Key Findings’, September 2015
 Office of the Chief Scientist 2014, Benchmarking Australian Science, Technology, Engineering and Mathematics. Australian Government, Canberra, p.73.
 Office of the Chief Scientist 2014, Benchmarking Australian Science, Technology, Engineering and Mathematics. Australian Government, Canberra, pp.26-32.
 Max Mason, ‘GDP masks economic problems, Saul Eslake says’, Australian Financial Review, 8 September 2015.
 Autor, David H., Frank Levy, and Richard J. Murnane. ‘The Skill Content of Recent Technological Change: An Empirical Exploration’, Quarterly Journal of Economics (2003): 1279-1333.
 David Autor, 2015, ‘Why Are There Still So Many Jobs? The History and Future of Workplace Automation’, Journal of Economic Perspectives 29, no. 3: 3-30.
 Goos, Maarten, Alan Manning, and Anna Salomons. ‘Explaining job polarization: Routine-biased technological change and offshoring.’ American Economic Review 104, no. 8 (2014): 2509-2526.
 Michael Coelli and Jeff Borland, 2015, ‘Job polarisation and earnings inequality in Australia’, Working Paper, University of Melbourne, Melbourne.
 Wilkins, Roger, and Mark Wooden. "Two decades of change: The Australian labour market, 1993–2013." Australian Economic Review 47, no. 4 (2014): 417-431.
 The chairman of the Australian Securities and Investments Commission gave a recent speech on how regulators are responding to this form of digital disruption. See Greg Medcraft, ‘Digital disruption and how regulators are responding’, Speech to The Regulators Panel, FINSIA (Sydney, Australia) 5 November 2015
 On one estimate, robo advisers will manage over $250 billion by 2019: https://www.investsmart.com.au/diversified-portfolios/robo-advice
 Julie Power, ‘Schools Find High Tech Era Down on Farm’, Sydney Morning Herald, 30 July 2015.
 Patrick Durkin, ‘Robots to Take 40pc of Low-Skilled Jobs’, Australian Financial Review, 2 October 2015.
 ‘A Bridge to the Future’, The Economist, 5 September 2015.
 Leigh, Andrew, and Chris Ryan. "Long-run trends in school productivity: Evidence from Australia." Education Finance and Policy, 6, no. 1 (2011): 105-135.
 Australia’s average PISA scores have been as follows. Reading: 528 (2000), 525 (2003), 513 (2006), 515 (2009) and 512 (2012). Maths: 524 (2003), 520 (2006), 514 (2009), 504 (2012). Science: 527 (2006), 527 (2009), 521 (2012).
 Annabel Hepworth, ‘Education a ‘national scandal’’, The Australian, 7 November 2015
 ACARA, ‘NAPLAN 2015 Results: Some Improvements, Time to Reflect’, Media Release, 1 August 2015.
 See for example Julia Gillard, ‘Ministerial Council for Education, Early Childhood Development and Youth Affairs - Communique’, 28 September 2009
 Lawyers data from Weisbrot, David --- "Recent Statistical Trends in Australian Legal Education"  LegEdRev 11; (1990-91) 2(1) Legal Education Review 219; Doctors data from ABS, ‘Health services: Medical practitioners’, Australian Social Trends, 2003, Cat No 4102.0; MP data from Joy McCann and Janet Wilson, 2012, ‘Representation of women in Australian parliaments’, Parliamentary Library.
 Leigh, Andrew, and Chris Ryan. "How and why has teacher quality changed in Australia?." Australian Economic Review 41, no. 2 (2008): 141-159.
 Justine Ferrari, ‘Brightest students dismiss teaching’, The Australian, 23 June 2014.
 An excerpt: ‘Finland, Finland, Finland / The country where I quite want to be / Your mountains so lofty / Your treetops so tall / Finland, Finland, Finland / Finland has it all’.
 For example, Finland was in the middle of the international pack in the 1962-67 IEA First International Mathematics Study, the 1967-73 IEA First International Science Study, the 1967-73 IEA Study of Reading Comprehension, and the 1977-81 IEA Second International Mathematics Study.
 Hanushek, E & Woessmann, L 2012, ‘Do better schools lead to more growth? Cognitive skills, economic outcomes, and causation’, Journal of Economic Growth, vol. 17, pp 267-321.
 Cornish, Selwyn. "Sir Leslie Melville: an Interview" Economic Record 69, no. 4 (1993): 437-457.