Opening the Black Box of Self-Employment: Identifying Alternative Work Arrangements in the United States
利用机器学习分析2003-2019年收入动态面板调查数据，识别自雇工作安排的多样性，揭示不同自雇形式的流行趋势、转换模式及从业者特征差异。
A substantial share of workers are self-employed, but there is a dearth of data on heterogeneity in these work arrangements. To address this gap, the authors identify the variety of self-employment work arrangements in novel data produced using machine learning, leveraging 2003–2019 Panel Study of Income Dynamics respondents’ narrative descriptions of their industry, type of work, and employer names. The authors examine trends in the prevalence and nature of these forms of self-employment, transitions across them, and who works in them. Findings show disparate trends in the prevalence of different work arrangements and in transitions across work arrangements that would otherwise be masked. Further results suggest that the informally self-employed are less likely to have business assets, engage in more routine and less abstract skills on their jobs, are less educated, are less likely to be male and non-Hispanic White, have less labor income, and have worse well-being.