Statistical theory and practice

The objective of this unit is to outline the general principles that underlie advanced methods of frequentist statistical inference. Building on maximum likelihood estimation (MLE), the unit provides a formal treatment of quasi-MLE, generalized method of moments (GMM), instrumental variable (IV) estimation, non-parametric methods and the bootstrap. The discussion is motivated by reference to econometric and statistical models and data, with insights provided into the theoretical properties and practical relevance of these alternative inferential techniques. The importance of asymptotic theory to the implementation of frequentist inference is highlighted, in a formal but intuitive way. Simulation experiments are used to reinforce the theoretical derivations.