However, technological shifts developed as growth in postsecondary education slowed and companies began to spend less to train their workers. “When technology, education and training go hand in hand, you get shared prosperity,” said Lawrence Katz, a Harvard worker. “Otherwise, you won’t.”
The increase in international trade encourages companies to adopt automation strategies. For example, companies worried about low-cost competition from Japan and later China invested in machines to replace workers.
Today, the next wave of technology is artificial intelligence. And Mr. Asemoglu and others say it can be used primarily to help workers, make them more productive, or relocate them.
Mr. Asemoglu, like some other economists, has changed his view of technology over time. In economic theory, technology is almost a magical ingredient that increases the size of both economic pies and makes nations richer. He remembers working on a textbook more than a decade ago that included standard theory. After a while, while doing more research, he came up with other ideas.
“That way of thinking is very restrictive,” he said. “I need to be more open-minded.”
Mr. Asemoglu is no enemy of technology. It notes that its innovations are needed to address society’s biggest challenges, such as climate change, and to deliver economic growth and rising living standards. His wife, Asuman Ozdagler, heads the Department of Electrical Engineering and Computer Science at MIT.
But as Mr. As Asemoglu digs deeper into economic and demographic data, the displacement effects of the technology become increasingly apparent. “They were more than I expected,” he said. “It makes me less optimistic about the future.”
Mr. Asemoglu estimates that half or more of the widening pay gap in recent decades stemmed from technology, published last year with his colleague, Boston University economist Pascal Restrepo. The conclusion was based on an analysis of demographic and occupational data that details the declining share of economic output that goes to workers as wages and increases costs on machinery and software.