Are Software Automation and Teleworkers Substitutes? Preliminary Evidence from Japan

Working Paper: CEPR ID: DP18406

Authors: Richard Baldwin; Toshihiro Okubo

Abstract: Digital technology is reshaping workplaces by enabling spatial separation of offices, known as telework, or remote intelligence (RI), and by facilitating automation of service sector tasks via artificial intelligence (AI). This paper is a first attempt to empirically investigate whether AI and RI are complements or substitutes in the service sector. It uses a worker-level panel of surveys collected from around 10,000 workers from pre-COVID-19 pandemic to late 2022, we find preliminary evidence that suggests that AI and RI are complements rather than substitutes. The evidence comes first from the positive correlation of investments in AI-promoting and RI-promoting software at the firm and worker level, and second from the positive correlation of workers' expectations regarding telework and software automation. The evidence is far from definitive but suggests that the complement-substitution question is a fruitful line for future research.

Keywords: Labor; Substitute; Complement; AI; RI; COVID-19; Telemigration; Globotics

JEL Codes: F0


Causal Claims Network Graph

Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.


Causal Claims

CauseEffect
Investments in AI-promoting software (C88)Investments in RI-promoting software (C88)
Expectations of telework (J29)Anticipation of work assisted by AI (D84)
Teleworkable occupations (J29)Susceptibility to automation (O31)
Investments in telework software (L86)Investments in automation software (O31)

Back to index