Today’s technological forces could dramatically enrich a few or, alternatively, lead to broad-based population-scale improvements in opportunity and the quality of life, writes Arun Sundararajan. It all depends on whether we choose the right policy priorities today.
The work that individuals perform to earn a living has always evolved over time. For example, at the turn of the 20th century, over 40 per cent of the U.S. workforce were engaged in farming; by the turn of the 21st century, this fraction was under 2 per cent. While creating vast new opportunities, this in-work evolution creates new needs for local and global mobility while raising new sustainability challenges.
Society is indeed accustomed to a gradual shift of the mix of occupations. But the pace of transition over the coming decades is poised for what will be, on historical scales, a dramatic digitally-enabled acceleration, driven by two major forces:
First, automation due to artificial intelligence and robotics technologies. Consider the retail checkout job. A few years ago, the cost-performance mix of the technologies needed to create an acceptable customer experience reached a level that enabled automation at scale. Over the coming two decades, similar technological thresholds will be reached in a range of other high-employment routine occupations, including factory work, providing customer service and driving a long-haul truck. The onset of Covid-19 has accelerated routine task automation.
Second, shifts to digitally-enabled platform business models. The primary impact of technology on the availability of work for humans does not stem from automated systems replacing or augmenting human workers. A more important source of change is the transformation of the business models fundamental to an industry, leading to a different mix of economic activities that rely less on human workers. For example, while many in-store retail jobs may have been replaced by automated checkout systems, many more have been lost to the growth of platform-based retailing. This shift has lowered the fraction of retail that occurs in stores. And since online retail does not involve any in-person interaction when checking out, there is a corresponding reduction in the availability of such work.
“safety nets today are designed for the 20th century workforce of salaried employees, and not for the emerging 21st century mix of work arrangements”
The onset of Covid-19 and the ensuing shelter-in-place actions have dramatically accelerated the adoption of platform-based business models for local retail and restaurant businesses. As economies rebound, there is also evidence that consumption in sectors like tourism and urban transportation is migrating away from the traditional and towards platform-based models like Airbnb, Uber and Didi.
Faced with this accelerating transition, what changes should we anticipate and how should we reshape policy?
First, we must expect a rapid rise in the non-employment labour force: people who derive their primary or supplemental income from freelance, gig, self-employment and other independent work arrangements. This rise was seeded by the emergence of digital platforms like YouTube that have created millions of content creators, and others like Facebook and Google whose advertising systems empower small businesses to emerge and reach their customer base.
Over the last decade, sharing economy platforms have further accelerated the pace at which freelancers and budding entrepreneurs can find customers and work directly through a digital platform. Tens of millions of entrepreneurs earn all or part of their income from Uber, Didi and Airbnb. On-demand and freelance labour platforms like Upwork, which manages relationships with over 100,000 clients, allow professionals to run independent businesses offering services ranging from administration and customer service to web development and accounting. Sector-specific professional labour platforms like Catalant for management consulting, and UpCounsel for law, are spawning thousands of full-time and part-time independent professionals. Food delivery platforms like Deliveroo, DoorDash, UberEats and Rappi have created a new generation of entrepreneurial restaurateurs who realise they do not need to take on the real estate costs of a storefront to serve their customers.
This rise in self-employment and entrepreneurship feels like a return to the past. One only need look back a century or so before realising that the dominance of salaried work is a relatively recent phenomenon. In fact, in 1900, about 50 per cent of the US workforce was self-employed, but this number shrank rapidly to 15 per cent by 1960.
However, these 60 years also coincided with the creation of government-led social safety nets around the world and, as a consequence, these safety nets today are designed for the 20th century workforce of salaried employees, and not for the emerging 21st century mix of work arrangements.
The key policy imperatives for the future of work therefore centre around addressing the needs that are uniquely associated with non-employment work. Two such needs are smoothing the greater short-term volatility that freelance income streams display, and refashioning government-sponsored risk management systems (like bankruptcy protection) to expect that an individual, rather than a large organisation, will bear much of the risk associated with shifts in demand or unexpected economic shocks.
Second, as automation dramatically alters human roles in the workplace, a larger and larger proportion of the workforce will have to switch occupations entirely mid-career. This is a far more significant transition than switching jobs within the same occupation, which is what we have become accustomed to over the last 50 years.
As a consequence, tens of millions of workers will need time away from work to reskill and transition. This necessitates shifting higher education investments away from the 20th century focus on early-career education, and towards a new 21st century focus on mid-career transition education. A forward-looking government will also create programmes to allow mid-career workers to bear the income shock associated with reskilling. Finally, the geographic regions where automation has the largest impact on job losses are almost always different from the regions where the new technologies are creating new work opportunities. Policy interventions that ease geographic mobility within a country are therefore imperative.
Contrary to what many believe, it is not the characteristics of a technology or technological progress itself that shape its societal impacts, and most saliently, whether the changes lead to greater or less inequality. Rather, these impacts are determined by the institutional, economic and political structures within which these changes occur. Today’s technological forces — artificial intelligence, robotics, digital platforms — could dramatically enrich a few or, alternatively, lead to broad-based population-scale improvements in opportunity and the quality of life. It all depends on whether we choose the right policy priorities today.