From Data to Decisions: Policy-Relevant Empirical Studies in Labor Economics

This dissertation studies how major long-term changes are reshaping labor markets and what they mean for labor market policies. It focuses on three broad forces: population aging, technological and data-driven changes, and the rise of non-standard forms of work. These topics are discussed in the four chapters of this dissertation. Each chapter presents empirical analyses that relate to the broader economic trends mentioned above. (1) The projected evolution of labor supply growth in response to demographic changes, (2) The consequences of domestic outsourcing for workers, specifically employment through staffing firms ('payrolling'); (3) the application of predictive models in student selection and the possibilities for affirmative action policies in such a setting; and (4) the use of forecasting models in predicting unemployment. While each chapter addresses a distinct topic in labor economics, the goal remains consistent: to generate empirical evidence that informs labor market policies. First, this dissertation examines demographic changes. As populations age and birth rates fall, worries arise about future labor shortages and slower economic growth. Using detailed Dutch data, the research shows that past labor force growth was driven mainly by population growth and increased participation rates of women. Future projections indicate a sharp slowdown in labor supply growth as population growth weakens and participation rates level off. These projection results imply that future economic growth cannot rely on an ever-expanding labor supply. Second, this dissertation analyzes changes in the way work is organized. The study presented in this dissertation examines payrolling, a form of domestic outsourcing in which workers are employed through staffing firms. The findings show that workers who move into these arrangements tend to experience lower job security, fewer permanent contracts, and weaker pension outcomes than their standard contract counterparts. This evidence highlights how flexible work arrangements can deepen the inequalities between secure and insecure jobs. Third, this dissertation explores how advances in data availability and computing power can affect decision-making. It shows how predictive models can be used to select students for studies that are subject to a limited number of places, in this case selection for admission to medical school. The research demonstrates that more diverse student populations using data-driven selection can be achieved with only small costs to efficiency, which can be measured using graduation rates. Finally, this dissertation considers economic forecasts, specifically for the unemployment rate. The final study of the dissertation shows that forecasting errors are not random but vary systematically over the business cycles. This chapter introduces a new way to measure economic turbulence and demonstrates that different forecasting methods perform differently depending on the degree of turbulence.