Jon Hilsenrath is a former senior writer at the WSJ, and author of ‘Yellen: The Trailblazing Economist Who Navigated an Era of Upheaval’
Tia Lee didn’t foresee a career when she first took a part-time job at a Macy’s make-up counter at the age of 17.
Looking back today at 58 — after becoming a professional make-up artist for television stars on Saturday Night Live, NBC’s Today show and elsewhere — she says her interpersonal skills made the difference. “When it comes to make-up, people are very needy,” Lee says. “It is intimate. You are touching somebody’s face.”
Her experience is a window into tectonic changes transforming the whole American workplace.
Technology and globalisation have driven decades of job upheaval in developed economies, akin to the decimation of work for skilled shoe cobblers and handloom weavers during the industrial era. While that was happening, the economy created millions of service jobs for people with social skills like Tia Lee’s. The rise of automated intelligence now threatens to shuffle the deck all over again.
To understand how work is changing, FT Alphaville revisited the outsourcing and trade debates of the early 2000s. Back then, many economists predicted vast destruction of US white-collar work due to the movement of services jobs to places such as India and the Philippines. We looked at the same database the economists used to make those predictions, the US Labor Department’s Occupational Employment and Wage Statistics, and examined what actually happened twenty years later.
Job loss did indeed turn out to be vast in many areas, perhaps even worse than the dismal scientists expected. Roughly half of all occupations that existed in the data set in 2000 experienced job loss over the subsequent two decades, with work for switchboard operators, filing clerks, transcriptionists, machine setters and many others collapsing. If you were in one of these occupations, your life was turned upside down and your ability to adapt put to the test.
At the same time, a different kind of work flourished: Tasks that required face-to-face contact and interpersonal skills, like make-up artists and manicurists. Skincare and professional make-up jobs grew from about 14,000 at the beginning of the new millennium to nearly 70,000 in 2023. Manicurist jobs grew from less than 30,000 to nearly 150,000.
Employment of meeting planners, personal financial planners, concierges, art directors, interior designers, occupational therapy assistants, coaches and athletic trainers all rose substantially, too. Project managers, private tutors, fundraisers and personal chefs became defined as whole new occupations that hadn’t been singled out as significant parts of the workforce before.
These occupational shifts are an essential feature and paradox of an economy built around computers and globalisation. The more work that machines have commandeered (or that was outsourced to cheap labour far away) the more the American economy created jobs for humans who interact directly with other humans, work that demanded something that computers don’t have and that wasn’t valued as prominently in the industrial era — the skill of social fluency.
The shifts of the past two decades polarised the job market, creating high-income work and low-income work while thinning out the middle class. It also drove social change, with women picking up work in fast-growing and high-paying occupations.
The result is what you could call the ‘Manicure Economy’, which is a window into how the next wave of technological change might affect our professional lives and the many political and social institutions built around them.
The ‘Manicure Economy’ in the age of AI
What will happen with the advent of artificial intelligence? There are few better authorities to ask than David Autor, a labour economics professor at the Massachusetts Institute of Technology, who has done more than any other economist in the past quarter-century to document the disruptive effects of trade and technology on the modern US workplace.
As he puts it:
I can’t think of technological change without thinking about how it relates to work. Work is our primary means of income distribution. You can have a world of marvellous technologies that many people don’t have the means to afford. Part of why the industrial era was so great is because it created shared prosperity. The concern about this new era that we have entered is whether we’ll have less of that or more of that.
In a 2016 paper that coined the phrase “China Shock,” Autor and co-authors David Dorn and Gordon Hanson showed how trade with China blew up work in scores of US manufacturing communities during the 2000s. The research undermined a core belief in economics that trade benefits all parties, a view that dates back to David Ricardo’s 1817 theory of comparative advantage.
In a series of other papers, Autor has detailed the phenomenon of job polarisation tied to technology, in which US employment grew for high-skill, high-wage jobs and low-skill, low-wage jobs, while middle-skill jobs declined.
At the core of his research was the theory that machines and global trade replaced rote tasks that could be coded and scripted, like punching holes in sheets of metal, routing telephone calls or transcribing doctor’s notes. Work that was left catered to a narrow group of people with expertise and advanced training, such as doctors, software engineers or college professors, and armies of people who could do hands-on service work with little training, like manicurists, coffee baristas or bartenders.
In keeping with Autor’s research, FTAV’s updated analysis of occupations data shows that many of the jobs that grew during the past two decades offered meagre pay, while some — including marketing managers, physician assistants and financial advisers — offered fast-growing six-figure incomes.
The slowest growing wages were in occupations in the middle of the income distribution, including die makers, millwrights and postal clerks.
More recently, Autor has been examining how AI might change the workplace. He is starting out — surprisingly — as an optimist who sees a future for middle-income workers not in spite of AI, but because of it. Autor argues that AI could put expertise in the hands of people without advanced training, creating work and pay gains for large numbers of less-skilled workers who missed out during the past few decades.
He cites one study of tech support workers by Stanford University professors Erik Brynjolfsson and Lindsey Raymond. It found that AI deployment shortened call durations, increased resolution rates and reduced turnover for low-skilled workers.
As the paper’s synopsis said, with FT Alphaville’s emphasis below:
New AI tools have the potential to change the way workers perform and learn, but little is known about their impacts on the job. In this paper, we study the staggered introduction of a generative AI-based conversational assistant using data from 5,179 customer support agents. Access to the tool increases productivity, as measured by issues resolved per hour, by 14% on average, including a 34% improvement for novice and low-skilled workers but with minimal impact on experienced and highly skilled workers. We provide suggestive evidence that the AI model disseminates the best practices of more able workers and helps newer workers move down the experience curve. In addition, we find that AI assistance improves customer sentiment, increases employee retention, and may lead to worker learning. Our results suggest that access to generative AI can increase productivity, with large heterogeneity in effects across workers.
Autor cites nurse practitioners as another potential example of a group that AI might help; with the assistance of technology, these professionals could take on decision-making now funnelled to medical doctors and reap better pay in the process.
“There is a future of work,” Autor says. At its core, he argues, is training and arming decision-makers, people with various kinds of expertise who make complex, on-the-spot choices about everything from how to treat a patient with chest pain to how to manage a customer who didn’t like how her steak was served.
However, it all depends on the choices people make now about how to implement technology. Autor points out that governments might choose to use it to spy on their people and companies could simply automate and replace more jobs. To rebuild a middle class, he argues, new technology has to be implemented with a sense of purpose and with humans in mind. “The future is not a prediction problem,” he says. “It is a design problem.”
This year Autor teamed up with Daron Acemoglu and Simon Johnson — two other star MIT economics professors — to launch an initiative aimed at driving the next wave of work changes before they happen, rather than watching and documenting their disruptive effects after the fact. As Autor puts it:
We have stopped being market fundamentalists and believing the market decides everything. We’re at the cusp of this transition and a lot of this stuff is up for grabs.
Rather than just writing research papers and holding academic seminars, he has been touring the West Coast, meeting executives at companies such as Google and Apple, and consulting with public officials to get these ideas implanted in their minds.
Jobsmageddon 3.0
Autor’s measured optimism about the future of work isn’t shared by others. For example, Elon Musk told the UK’s prime minister last year that “there will come a point where no job is needed”.
In such a world, Musk has argued, governments will need to create a system of universal basic income to ensure that households have the means to consume. This will obviously lead to profound and divisive political questions about the role of government in the economy.
Such dire predictions aren’t new. Remember the word “Luddite?” They were 18th-century British weavers and textile workers who objected to the use of mechanised looms and knitting frames, with some breaking into factories and smashed machines. They called themselves “Luddites” after Ned Ludd, a young apprentice who became the figurehead of the movement. The image at the top of this post – showing a NYT headline from 1928 – shows how workers remained angst-ridden about factory machinery over a century after the Luddites first tried to prevent technological change from disrupting their lives.
The current angst is also reminiscent of two decades ago, when fears of the effects of outsourcing on white-collar jobs gripped the economics profession. At the time, Princeton professor Alan Blinder estimated that more than a quarter of all US jobs were prone to outsourcing within a decade or two.
Some occupations that he saw as threatened — such as telemarketers — were indeed brutalised. Jobs in that category dropped from nearly half a million in 2000 to 81,000 by 2023. As Blinder says:
That old [research] is strangely relevant today, but in a different way than I was thinking about then — because a major determinant of whether a job was ‘offshorable’ then was how important face-to-face (as opposed to electronic) contact was.
Nowadays, we are asking similar questions about which jobs could be replaced by AI-powered machines. In this case, it may not matter whether the machine is onshore or offshore. In either case, there may be net job loss. I say ‘may be’ because new technologies always create new jobs even as they destroy old ones.
Yet some of the occupations that Blinder and others saw as vulnerable in those earlier studies instead grew robustly. For example, employment of customer service representatives rose from 1.9mn in 2000 to 2.9mn in 2023.
Harvard economics professor David Deming found a vital missing link. He had been wondering why early childhood education programmes led to lasting economic gains for people who took part in them. Perhaps the simple lessons of pre-school — such as lessons to share and play well with others — might matter as much or more than learning to count and spell in an advanced economy?
He discovered that he was on to something. Jobs requiring high levels of social interaction were growing robustly, Deming’s research found. Pay increases were especially large for people with combinations of math and social aptitude. Teamwork, Deming found, was an essential new ingredient to work success. Research in other countries including Sweden and the UK found similar high returns to people with strong social skills.
As machines and outsourcing replaced routine, rules-bound work, the marketplace was increasingly demanding and delivering work that involved not only physical presence, but also the unconscious formulations that allow humans to read and adapt to other humans as customers and co-workers.
One papers sums up his findings like this:
A growing body of work emphasises the importance of “non-cognitive” or “soft” skills like patience, self-control, conscientiousness, teamwork, and critical thinking. While such skills are clearly important, the very terms “soft” and “non-cognitive” reveal our lack of understanding about what these skills are and how to measure or develop them. In my view, the appropriate term for capacities like problem-solving, critical thinking, and teamwork is higher-order skills.
Shifting American spending habits helped to drive these changes. As goods became mass produced at factories, the US economy became more oriented towards spending on services. Since 1980, US spending on services has risen from about half of all spending to two-thirds, with increasing shares going to health and education, and in recent years to dining out and hotels. “A lot of that is the result of an advanced, prosperous economy,” Deming says.
Against this backdrop, customer service jobs grew because representatives provided a human touch that wasn’t easily replaced by machines or call centre operators on distant shores. Some companies — including Apple, Delta Air Lines and AT&T — tried offshoring customer support jobs overseas and then brought some back to the US when customers complained about language and cultural barriers.
Customer service jobs may again be threatened, this time by AI bots that can chat with humans online, though the Stanford study suggests AI might also arm humans — including low-skilled workers — to handle the most complex customer service problems more effectively.
As Deming says:
Non-routine interaction is at the heart of the human advantage over machines. The whole point is the relationship. I don’t want a robot to be my coach.
This is already happening. Jobs for coaches and scouts are among those that increased rapidly over the past two decades, from less than 70,000 in the year 2000 to nearly a quarter of a million by 2023. Ironically, jobs for human resource specialists grew from less than 200,000 in the year 2000 to nearly 900,000 by 2023.
Julia Pollak, chief economist at ZipRecruiter, says the work of human resources specialists is no longer confined to rote administrative tasks. It has broadened to managing widening arrays of personnel problems. In keeping with that shift, the way that the labour department describes the tasks of human resource jobs has changed over two decades, as has its classification of many other jobs.
In 1998, the department’s category for HR hiring specialists said simply that these individuals “recruit and replace workers”. By 2018, it had a much more elaborate and specialised description of the work: “Recruit, screen, interview, or place individuals within an organisation. May perform other activities in multiple human resources areas. Excludes compensation, benefits, and job analysis specialists and training and development specialists.”
At the same time, work that didn’t even exist before, or that hadn’t been thought of as its own individual occupation, has been born. This includes jobs including web developers, solar photovoltaic installers, wind turbine technicians and genetic counsellors. One ancient occupation — acupuncturist — became large enough to get its own category.
Sheconomy vs Heconomy
Alisha Rimondo was thinking about becoming a lawyer before dropping out of college in the 1990s and making her career in the manicure industry, just as the industry was entering a boom.
New technology, in the form of gel polish, made manicures more lasting. Immigration introduced cheap labour, which drove down costs. Tastes evolved, with a growing number of women — and some men — seeing manicures as a new form of self-expression. Between 2012 and 2022 the number of multi-employee US nail salons doubled, according to the labour department.
“More demand from people led to more technicians being needed,” says Rimondo, who ran salons, wrote textbooks, marketed manicure products overseas, and even became a judge in an annual “Nailympia,” a competition for elite manicure designers. Her mother, an Oklahoma chemist in the oil and gas industry, was unhappy when she left school. However, within a few years Rimondo was earning more than her mother. “Everything doubled,” she says.
In a workplace that increasingly caters to soft skills, men seem to lack advantages that helped them on factory floors during the industrial era, like sheer brute force, with potentially vast social and political implications.
In a 2018 working paper titled The “End of Men” and Rise of Women in the High-Skilled Labor Market, economists Guido Matias Cortes, Nir Jaimovich and Henry Siu found that between 1980 and 2016, the increased importance of social tasks in work helped to explain a growing share of women in high-paying occupations relative to males.
My own research showed that women made up half or more of the workforce in the fastest-growing occupations between 2000 and 2023, including health, legal, finance, education and arts, while their share has grown in management and sciences. In legal occupations, for example, women’s share of work grew from 46 per cent to 52 per cent over two decades, and in management occupations women’s share grew from 37 per cent to 42 per cent.
Meanwhile, men are over-represented in production occupations decimated by automation and globalisation. Women also had a large share of jobs in some low-pay occupations that lost work, such as administrative assistants, but analysts see the work landscape tilting in their favour.
The McKinsey Global Institute estimates that as many as 28 per cent of men in advanced economies could need to switch occupations by 2030 because of technological change, compared with 24 per cent of women.
Similarly, Brookings Institution scholars Mark Muro, Robert Maxim and Jacob Whiton estimate that 24 per cent of male workers hold jobs that are at potential high risk from automation compared with 17 per cent of women, because women work in occupations less easily replaced, including health services and personal services.
The changing nature of work has in turn played a role in political and social disruption. One of Autor’s papers suggested that communities where white men were hit hardest by the China trade shock of the 2000s became more politically conservative and oriented towards Republicans. In addition, marriage and fertility declined in those communities, while the share of unwed mothers rose along with the incidence of premature death among men.
Autor sums it up with understatement characteristic of an economist:
The past four decades have been quite a mixed picture for the US labour market and for US democracy.
How can you prepare yourself for the next work onslaught? If the past is any guide you don’t necessarily need to brush up on your computer skills. You need to get better at dealing with other humans.