Empirical studies of labor markets increasingly model hiring, separation, job creation, job destruction, search, and wage determination in a manner that dictates the use of longitudinally linked employer-employee data. There is an important structural connection between dynamic random graph models and the statistical analyses of such data. Bipartite graph theory exposes this relationship and permits applied models to exploit more fully the information in the linked data. Individuals and employers are modeled as network nodes realized from evolving populations. These are the disjoint node sets of the bipartite graph. The employment relation is modeled as an undirected link between elements of these node sets. Identification of statistical models on this graph is determined by properties of the adjacency matrix. Dynamic models are identified by properties of the transition matrices for the evolution of the adjacency matrix.
This course will provide an overview of labor market theory that motivates the use of linked employer-employee data. Using prototypical wage determination models, the connection to dynamic random graph models will be illustrated. Statistical models that exploit this linkage will be examined. Hiring, separation, job creation and job destruction models will also be illustrated using the graph modeling techniques. Other problems in labor economics, specifically job duration and job-to-job transitions will also be studied.