The Asian Development Bank has issued a lengthy report on technology and jobs. And it’s a pretty upbeat one, as described by the Financial Times:

The ADB’s analysis of 12 developing Asian economies between 2005 and 2015 found that rising demand had more than compensated for jobs lost to automation. The adoption of new technologies, such as modern machine tools and computer systems in factories and offices, had stimulated higher productivity and economic growth. That transformation, it estimated, had created 134m new jobs, compared with the 101m jobs lost to technology.

I also wanted to highlight a couple of graphics, which try to encapsulate how researchers look at the tech-jobs issue. About this first graphic:

To illustrate, Figure 2.1.9 shws that occupations with a larger share of routine tasks are more likely to be automated, while those with a lower share are less likely. Workers in occupations in quadrant 4 (researchers and managers) are safe because the majority of their tasks are difficult to automate and new technology augments the value of their labor. In contrast, workers in quadrants 1 and 2 hold jobs with mostly routine tasks, including cognitive routine jobs (accountants and bank tellers) and manual routine jobs (sewing machine operators and assembly line workers). These jobs are at risk of displacement by labor-saving technology. Manual and nonroutine jobs in quadrant 3 (cook and hairdresser) are not yet heavily affected by laborsaving technology.

This second graphic tries to illustrate some of the research of MIT’s Daron Acemoglu and Boston University’s Pascual Restrepo that I have been blogging about in recent months. In particular, it looks at the main channels through which technology may affect jobs and earnings.

 (i) Displacement effect. Because robots and computers are good at routine tasks, demand will fall for jobs comprising mainly routine tasks. A manual worker in an industrial warehouse whose job is to fetch products from shelves, for example, is likely to be displaced, perhaps by Kiva system robots, which can traverse large floor spaces to find products much faster than humans. Interestingly, though, displacement is bounded by tasks that humans accomplish effortlessly but computer programmers struggle to code into routines. Polanyi’s Paradox recognizes that we know more than we can tell (Autor 2015; Polanyi 1966).

(ii) Productivity effect. Sometimes called a scale effect, it is when automation improves productivity and lowers production costs. Under normal conditions, this lowers the price of goods and services, which raises demand for them. As industrial robots become more sophisticated and widely used in production lines in Asia and the Pacific, for example, the cost of producing cars could go down, pushing down prices and spurring increased demand for cars. To the extent that increased demand requires hiring more workers, it could offset the displacement effect from automation.

(iii) Reinstatement effect. Automation can spawn new labor-intensive tasks and jobs, raising demand for labor. New job categories could emerge as AI is introduced into production, for example, or when a more sophisticated industrial robot is introduced on a factory floor and needs programming or tending. Cross-industry effects. Adopting new technology in one industry has an impact on productivity and jobs in other industries.

There are two main channels through which cross-industry effects change labor demand. 

(i) Spillover effect. As one industry adopts new technology, positive spillover affects other industries in at least three ways. First, firms in downstream industries benefit from cheaper and/or better-quality inputs, while firms in upstream industries benefit if the output of the automating industry expands. Second, other industries learn the benefits of adopting the new technology. Third, workers with new skills and knowledge move between industries, spreading technological know-how.

(ii) Income effect. When technology complements labor, workers’ higher incomes create positive spillover on other industries through increased demand for goods and services. A software developer whose income has increased thanks to complementarity between automation and human labor, for example, may want to buy a bigger car, a faster computer, better health care, more vacations, or other leisure services.