This book attempts to sum up reviews and discussions about causal inference from the statistical and econometric literatures, describing and comparing these two main approaches. Particular attention is on dynamic models that allow to control for unobserved heterogeneity when longitudinal data are available. The application concern the labour market where in the last years, the average length of employment relationships has significantly been reduced with a consequent higher rate of total turnover. In order to define the characteristics of a good job a new definition of stable job is...