If tech is driving the ‘productivity bandwagon’, it’s time to hit the brakes

Technology is everywhere and always an unalloyed good. New technologies ultimately create better jobs and more broadly based prosperity. So goes the conventional economic wisdom. But what if it wasn’t true? What if technology had been used — in lieu of strong political and institutional restraints — to put more money in the hands of elites throughout history? 

That’s the starting point of Power and Progress, an upcoming book by MIT economists Daron Acemoglu and Simon Johnson, to be published next month. It explores several moments over the last millennium when technology led to the opposite of shared prosperity: agricultural improvements that created almost no benefits for peasants; advances in ship design that allowed the slave trade to grow; and industrial factories that took flexible craft work out of the home and put it under the control of managers who increased working hours and decreased pay. It also addresses more recent developments, such as automation used to micromanage labour — and the coming revolution in AI that may disrupt us all.

These economists are hardly technophobes. It’s probably impossible to be one at MIT, a centre of American innovation. But the two academics take a different approach to the productivity gains of technology and how they get distributed compared with most of their peers. Neoclassical economic theory holds that technological progress always increases average wages. And even if it raises inequality, it ultimately lifts wages at the bottom of the income distribution. Acemoglu and Johnson refer to this idea as the “productivity bandwagon.” 

But the pair show that automation — perhaps the most important technological advance since the industrial era — isn’t about increasing labour productivity but rather replacing it. Automation doesn’t necessarily reduce wages if there are incentives or requirements (on the part of unions or government) that force the retraining of displaced workers and the creation of new jobs for them. But this isn’t always the case. If new jobs and tasks aren’t actively created, then automation can end up decreasing jobs and wages, even as it increases productivity and returns to capital. 

This is, of course, largely where we have been over the last several decades, as the economic pressures on managers to hoard capital and treat workers as a cost rather than an asset on the balance sheet have grown.

Things didn’t always work that way. Consider the rise of electric power in the 19th century and the effect that this had on labour. Jobs for engineers and white collar managers increased, as they used the innovation to remake the shape of factories and create new and more efficient jobs for workers. This process continued for decades, helped along by New Deal legislation that encouraged collective bargaining and reduced corporate concentration (and thus political power), as well as by strong unions that made worker retraining part of the social compact. By the 1960s, the income share of the top 1 per cent of the population had fallen to 13 per cent, down from 22 per cent in the 1920s. Average wages grew as fast if not faster than productivity.

From the 1970s onwards, that link started to break, in large part because of the decline of unions, shifts in antitrust policy, accounting changes that incentivised debt over productive capital expenditure in things like training, and a general dovetailing of technological disruption and outsourcing. All this meant that even as American workers were becoming more productive, they weren’t sharing in the fruits of that productivity growth.

The result was the implementation of what the authors call “so-so automation,” such as worker-tracking software or call centre bots, which aren’t actually that much more productive than humans, if indeed they are at all (think about how long it takes software to solve a customer relations problem versus a human). Such “innovation” mainly just decreases costs for employers. 

We are now at a turning point in the story of technology. Even the titans of Silicon Valley — people like Elon Musk and Apple co-founder Steve Wozniak — are calling for a slowdown in the rollout of AI, so that its implications can be better studied. Google and Microsoft are telling us there’s nothing to worry about. All this reflects the power of persuasion by influential figures, something the authors explore in detail. 

Throughout history, major entrepreneurs in technological innovation, from Ferdinand de Lesseps (who was responsible for the Panama Canal building debacle) to the titans of so-called surveillance capitalism who pushed rules that allowed them to mine and profit from our personal data, have used power and influence to set the narrative around technology, which then takes on a life of its own.

We cannot allow that to happen now. Technology has created shared prosperity only when appropriate democratic guardrails have been in place to make sure that it does so. AI poses threats to both democracy and to jobs across all income bands. The result may be quite dystopian. Unions and government alike must act to make sure that this latest ride on the productivity bandwagon doesn’t end in tears.

rana.foroohar@ft.com