A Statistical Investigation on Workforce Automation

Abstract

A handful of highly automated businesses are continuing to profit handsomely in the midst of the COVID-19 pandemic. This paper takes interest in the underlying relationship between an occupation’s susceptibility to automation and variables of economic interest, controlling for a number of factors to identify high-risk characteristics. Using datasets from the Bureau of Labor Statistics and the Department of Labor’s Employment and Training Administration’s Occupational Information Network, we find the degree of computerized tasks and wage level to be statistically significant variables. Other predictors, such as the programming skill of workers, are borderline statistically significant, with p-values slightly exceeding 0.10. Overall, we determine that occupational susceptibility to automation is negatively correlated with educational attainment (in the absence of related predictors), wage, and employment growth rates from 2010-2019. However, we find it is positively correlated with real wage growth rates over the same period and again during the pandemic recession. The methods and final results vary from that of the existing literature, most likely due to differences in variable and model selection. This study accomplishes the following: (1) forms a probabilistic model of workforce automation that is accessible to both economists and public policy makers; (2) provides economic interpretations to our model’s predictions with additional parametric models.