Quantitative insights into labor dynamics[Analyse quantitative de la dynamique du marché du travail]

This dissertation investigates wage inequality and labor market sorting by tracing an analytical path from institutional boundaries to the structural econometrics of worker-firm-occupation networks. Chapter 1 examines the public-private pay gap using dynamic lifetime earnings rather than static hourly wages. Analyzing French administrative data, it reveals that institutional structures create distinct sectoral winners and losers based on gender and education, exposing profound unobserved heterogeneity. To capture this complexity, Chapter 2 shifts to a multidimensional framework, implementing a Classification Expectation-Maximization (CEM) algorithm to jointly estimate three-way sorting among workers, firms, and occupations. By evaluating jobs as triplets, the analysis demonstrates that occupation is the primary driver of both wage and lifetime-value variance, and that worker mobility is overwhelmingly driven by immediate wage gains rather than option value. Finally, Chapter 3 evaluates the econometric foundations of bipartite worker-firm models (AKM estimators). Utilizing spectral graph theory, it demonstrates that additive variance decompositions are highly sensitive to network sparsity. It identifies the average node degree as a crucial diagnostic for spectral collapse and noise amplification, offering a framework to assess the reliability of empirical sorting estimates. Together, these chapters advance the labor economics literature by demonstrating that comprehensively mapping wage dispersion requires integrating institutional dynamics, multidimensional career mobility, and robust, network-aware econometrics.