Source: Dechezleprêtre, et al., “Induced Automation: Evidence from Firm-level Patent Data,” University of Zurich Working Paper Series, April 2021.
Commentary: Automation is one of the best responses firms may take when the cost of labor increases. They can offset the cost of higher wages in their supply chains by increasing investment in labor-saving automation technologies, which in turn lowers their demand for labor. While this rationale is common among economists mindful of automation, empirical data on the relationship between low-skill wages and innovations in automation have previously been limited. New research by economists at the University of Zurich tests the hypothesis that rising low-skill wage is a driver of automation by examining firm-level data of automation patents.
Their research identified automation patents in machinery capable of substituting worker activity by compiling the textual data of patents maintained by European Patent Office databases and computing the frequency of patents containing keyword indicators of automation. Text-as-data analysis in patent research produces a new and more accurate proxy for measuring changes in automation. By modeling the relationship between wages and automation over a span of 12 years using this proxy, the study found that a 1 percent increase in low-skill wages was associated with a 2 to 3 percent rise in automation patents filed in the residing country where that wage hike took place. Second, an increase of 1 percent in the low-skill wage is associated with 4 to 5 percent more automation patents in foreign countries. Automation innovations in machinery are more elastic than other innovations and the empirical findings of this study echo long-held economic expectations about firm responses. Low-skill labor reflects a greater feasibility for advancements in machinery to replicate human tasks and automation can function empirically as a cost-saving substitute for production.
