Economists explore the consequences of steering technological progress

SINCE THE ancient Greeks, at least, people have recognised that civilisational progress tends to create havoc as well as opportunity. Economists have had little time for such concerns. To them, technological progress is the wellspring of long-run growth, and the only interesting question is how best to coax more innovation out of the system. But in the face of looming social challenges, from climate change to inequality, some are now asking whether, when it comes to innovation, what sort is as relevant as how much.

Early models of growth did not explain technological progress at all, treating it rather like manna from heaven. In the 1980s some economists worked to build endogenous-growth models that said where innovation came from. They explained it as the consequence of investment in research and development, increases in the stock of human capital, or the (temporary) extra profits that can be reaped by firms with new technologies. Other economists have focused more on data than on theory. “Who Becomes an Inventor in America? The Importance of Exposure to Innovation”, a paper published in 2018 in the Quarterly Journal of Economics, identifies factors that seem to encourage young people to become innovators. Children who grow up where innovation rates are high, for instance, are more likely to become inventors themselves.

Research has also made clear, however, that technological discovery is not linear, but veers about depending on economic conditions. Some economic historians reckon that early industrialisation was motivated by a desire to replace scarce resources, such as skilled labour, with abundant ones, such as unskilled labour and coal. Early inventors were not simply discovering nature’s truths one by one, in other words, but trying to solve specific problems. Work on such technological bias blossomed in the 1990s as economists sought to explain why the wage premium earned by college graduates kept rising even as the supply of graduates increased. The answer, some reckoned, was that technological change in the 20th century was “skill-biased”, boosting the productivity of workers with degrees, but not of others.

In a paper published in 2001, Daron Acemoglu of the Massachusetts Institute of Technology collected these strands in a model of “directed technical change”. Technological progress, he suggested, is influenced by the relative scarcity of factors such as labour and capital; by how easily one factor can be substituted for another; and by the path of past innovation. Research on a particular technology may reduce the cost of developing complementary innovations in future. Directed technical change is fascinating to contemplate because it allows for alternative technological futures: worlds in which firms wring every efficiency from Zeppelins and pneumatic tubes, rather than from internal-combustion engines and Twitter. If the direction of progress is not set in stone, policy choices could lead an economy down one technological path rather than another. That raises an immediate question: if innovation can be steered, should it be, and if so, how?

Since 2000, published work on directed technical change has focused largely on environmental challenges. Path dependence means that research on fossil-fuel technologies can often be more fertile than research on cleaner alternatives. There are more experts in the relevant disciplines, better-funded research labs and an established complementary economic infrastructure. Efficient decarbonisation might thus require subsidies for clean-energy research, as well as a carbon price. Indeed, efforts to slow global warming represent a massive attempt to realise one technological future—a zero-carbon version—rather than another.

Why stop there? Some futurists, and a few economists, worry that rapid progress in artificial intelligence could lead to mass displacement of labour and social crisis. But in a recent paper Anton Korinek of the University of Virginia notes that not all uses of AI are alike. Clever machines could indeed replace human workers—or might instead be engineered to assist human labour: to help people navigate complicated processes or take difficult decisions. Private firms, focused on their bottom lines rather than the potential knock-on effects of their investment decisions, might be indifferent between the two approaches in the absence of a government nudge, just as polluting firms tend not to worry about the social costs of environmental harm unless made to do so by governments. In a working paper co-written with Joseph Stiglitz, a Nobel laureate in economics, Mr Korinek concludes that directing technical change to favour labour-assisting rather than labour-displacing forms of AI could be a second-best way to manage progress, if governments cannot sufficiently redistribute the gains from automation from winners to losers. This may sound far-fetched, but policy proposals such as Bill Gates’s suggestion that robots should be taxed to slow the pace of automation represent steps toward a more micromanaged technological future.

Oh, the humanity

Environmental policies aside, such steps seem premature. A more sophisticated view of technological progress is to be welcomed. But economics lacks the tools, at least for now, to judge which technological path is preferable. The world is too complex to allow economists to compare hypothetical technological futures: to know whether a Zeppelin-based society would operate more efficiently overall than a car-based one. Economists cannot know what surprises lie down one innovation path rather than another.

And questions of technology are not solely, or even mostly, about efficiency. Many are ethical. Innovations with overwhelming productivity advantages could prove devastating to social trust or equity. In the face of radical technological change—in AI, robotics and genetic engineering—societies will inevitably argue over which technological paths should be explored. Economists’ views belong in these conversations—provided they are crafted with humility and care. ■

Explore our A to Z of economics