Thesis title: Essays in Applied Labor Economics: Causal Inference on Gender, Time Allocation, and Fertility at the Retirement Threshold
The ageing of the population and the sustainability of welfare systems pose crucial challenges for European economies, making retirement a pivotal event at both the individual and policy levels. The transition from work to pension not only redefines an individual's life but also triggers profound transformations in family dynamics, with significant gender and intergenerational implications.
The challenge this dissertation seeks to tackle is twofold. First, it aims to estimate the causal effect of retirement on two key dimensions of family behaviour: time allocation and fertility choices. Specifically, it analyses how men and women reorganise their time between domestic work and leisure after retirement, and how the newly found time availability of grandparents influences the reproductive decisions of their children. To credibly isolate these causal effects, the dissertation adopts a Regression Discontinuity Design (RDD), using statutory pension eligibility ages as a source of exogenous variation. Second, the dissertation contributes to the advancement of this methodology by developing tools that integrate Machine Learning techniques to enhance the precision and robustness of estimates in the presence of complex data.
The manuscript is structured as a sequence of three chapters, all connected by the adoption of the RDD framework to analyse socio-economic phenomena. Chapter 1 applies a fuzzy RDD to assess the causal effect of retirement on time allocation between men and women in Italy, examining the gender implications for healthy ageing. Chapter 2 presents a purely methodological contribution, developing a new Stata command, rdlasso, which implements a Lasso-based procedure for selecting high-dimensional covariates in both sharp and fuzzy RDD settings. Finally, Chapter 3 extends the analysis to a pan-European context, using the RDD framework to study the impact of grandparental retirement on the fertility of their adult children, testing the heterogeneity of this causal link across different welfare systems.
In chronological terms, each chapter of this doctoral thesis represents a modest advance over the previous. While Chapter 1 rigorously applies the RDD methodology to a critical social policy question, Chapter 2 directly addresses a methodological challenge by creating a practical tool to enhance the technique itself. Chapter 3, in turn, synthesises the previous two approaches by applying an RDD methodology, potentially enhanced by the techniques discussed in the second chapter, to a complex, large-scale causal analysis, thereby demonstrating the flexibility and robustness of the proposed methodological framework.