Year of Graduation
“Mover-stayer” binary choice model for panel data
School of Applied Mathematics and Information Science
The goal of the graduation project is to describe binary choice models where considered observations are divided into two latent clusters. Each cluster has different logistic equation representing the relation between dependent variables and an independent variable. The graduation project is based on the detailed review of the domain of binary choice models in logistic regression as well as closely related Split population model for cross-section data. In the current paper a verbal and formalized description of the model “mover-stayer” for panel data is proposed. The model is programmed in statistical software package “Stata” with the help of advanced algorithms of maximum likelihood functions estimation by the help of “ml” functions. Moreover, different versions of model panel data were generated to analyze the proposed model’s characteristics using Monte-Carlo simulations. What is more, the model was applied with the real dataset that was taken from the research of the problem of female employment conducted by Russian economy and health monitoring. The calculated results are provided and thoroughly discussed in the paper. In the future the developed program may be successfully applied to the analysis of any other data within the limits of the designed model.