Abstract

To examine how technological progress can affect labor demand, I analyzed the employment effects of technological progress by industry using two types of firm-level data: financial data and annual business reports. Since the characteristics of technology through technology-related financial variables may be limited, the effects of technological progress were analyzed by considering the heterogeneity of technology through the contents of annual business reports in addition to financial variables.
First, technological progress based on financial variables such as intangible assets and R&D expenses significantly increased employment in the overall and in many individual industries, and generally, the larger the firm, the greater the employment increase effect.
On the other hand, the three technologies estimated through annual reports showed differences in employment effects depending on their characteristics.
In the case of technological progress related to R&D and patents (tech-words 1), similar to financial variables, there was an employment increase effect in a number of industries, while technological progress related to robots, IT and automation (tech-words 2) had an employment reduction effect in some industries. Finally, AI-related technologies (tech-words 3) had an overall employment reduction effect, mainly in the manufacturing and construction industries, and also showed significant differences in employment reduction effects by firm size in some industries in the service industry. These results suggest that the direction and magnitude of the employment effects may vary depending on the heterogeneous characteristics of technology.
Unlike the results based on financial information, the employment effects of technological progress based on text information showed significant and large effects in small-sized firms in many industries compared to large ones, and differences in results could be found depending on the measurement of technology and characteristics of firm. Looking at the combined effect of the three technologies based on significance, I also found differences by industry, with manufacturing showing an overall decrease in employment and services industry showing an increase in employment due to technological progress.
However, these results should be interpreted with caution, as they have the semantic limitations of technologies represented by simple keywords and exclude the dynamic employment effects of technological shocks.