Double Whammy? The Impact of Trade and Automation on High-Skilled Jobs

As Artificial Intelligence (AI) capabilities for language, speech and image recognition pass human levels, AI-enabled software can perform white collar tasks previously done exclusively by high-skilled humans. Furthermore, white collar jobs in rich countries could face a double whammy. Not only may AI-enabled automation transform or replace such jobs, but digitized tasks can also be traded over electronic networks such that an engineering job in, say Germany, can be performed remotely in India. For the double whammy scenario to come to bear, AI-enabled automation must be broadly adopted, and professional services widely traded. Furthermore, imported services and AI must replace local professionals faster than new jobs are created.

So far, fear of disruption and mass unemployment in white collar jobs seems not to be supported by facts. To the contrary, despite a substantial increase in imports of professional services and high exposure to AI, employment in professional services has grown steadily in the OECD area, both in absolute and relative terms. In all OECD countries for which sufficiently detailed and updated data are available, employment in professional services has grown faster than overall employment (Figure 1). Moreover, imports of professional services have grown faster still in these countries (except for Greece, which went through a deep economic crisis during the period recorded).

Figure 1. Employment % change from 2010 to 2019

Source: OECD Stan database.

Note: Total refers to change in total employment. Professional services refer to ISIC rev 4 sector D69-75. The figure includes the countries for which data was available for both 2010 and 2019 for professional services. The countries are (from left to right) Canada, Costa Rica, Greece, Ireland, Iceland, Turkey, and the US

Figure 2. Imports of business services by category, % change from 2010 to 2019

Source: OECD Trade in Services database.

Note: EBOPS categories SJ: Other business services; SJ2: Professional and management consulting services; SJ3 Technical, trade-related, and other business services.

Engineering firms create industrial applications from technology developed in labs, playing an important role in AI technology adoption in manufacturing. In an earlier blog we explained that the uptake of AI in manufacturing, like other technologies before it, is S-shaped. It starts with a few highly productive and innovative firms, picks up speed over time, and levels off when most firms have adopted the new technology. Both technology immaturity and lack of adequate skills hold back adoption at the early phase. Engineering firms in turn, follow a boom-and-bust cycle in creating industrial applications for which the market may not always be ready. Currently, AI is at the slow bottom segment of the S-curve.

In a new simulation study, we add the international trade dimension to the analysis and investigate whether skilled workers, in our case engineers, face the double whammy of being squeezed from both import competition and automation. The short answer is no. In the short to medium term skills shortages are more likely than idle professionals. In the long term, however (about 20 years in our simulations), demand for professionals in the engineering-manufacturing complex could decline substantially due to automation.

Our findings are summarized as follows:

• Imports of high-skilled services from low-cost countries postpone automation in rich countries. Manufacturers differ in productivity and the cost advantage of automation is higher for the most productive firms. Manufacturers will automate when it pays off for them. Imports of low-cost services, e.g., through remote offshoring from India, reduce the cost advantage of automation and raises the productivity threshold for switching technology.
• Exports of AI-enabled software bring AI-adoption in manufacturing forward in the software-exporting country. The software exporters harvest data from their customers – and use the data for updating and improving the software, to the benefit of local manufacturers as well. Thus, exports – i.e., local engineering firms engaging in cross-border licensing of software – lower the productivity threshold for local manufacturers to switch technology.
• Market integration accelerates the adoption of AI-enabled automation. The reason is that there are network effects in adopting technology. These are weaker across than within borders. Therefore, large countries adopt AI faster than small countries, but trade narrows the difference.

There are a host of policy parameters that influence these outcomes and their magnitude. Any trade policy measure that raises trade costs curbs the trade channel of technology adoption.

Occupational licensing, if it raises the wage premium of engineers, contributes to faster AI adoption in manufacturing at home. It also raises entry barriers for foreign engineering services providers. Furthermore, if the reservation of a set of tasks to licensed professionals also applies to automation software, it may affect the feasibility of cross-border software licensing.

Restrictions on cross-border dataflows limit or nullify the impact of cross-border licensing on AI adoption in manufacturing at home.

Finally, cross-border enforcement of intellectual property for software affects the cost and feasibility of cross-border licensing.

To conclude, while technology spillovers through the flow of data and people encourage the development of AI applications, imports of low-cost services discourage its uptake. On balance, high-skilled jobs exposed to AI-driven automation and import competition are most likely safe for now.

 

References

Felten, E. W., Raj, M., & Seamans, R. (2019). The occupational impact of artificial intelligence: Labor, skills, and polarization. NYU Stern School of Business.

Klügl, F. and H. Kyvik Nordås (2021), Is Artificial Intelligence coming for your job? CEP Blog 13 June

Klügl, F. and H. Kyvik Nordås (2021). AI-enabled Automation, Trade and the Future of Engineering Services. Working Paper 16/2021, Örebro University, School of Business.