Topics in dynamic model analysis. Advanced matrix methods and unit-root econometrics representation theorems. (Q2569529): Difference between revisions

Our the past two decades, econometric theory gave a preferential treatment to the role of time series techniques, while a special algebraic toolkit supported the (co)integration theory advancements. The basic methods of (dynamic) simultaneous equation models (SEM), time series econometrics for the analysis of economic data with vector autoregressive (VAR) models, and the analytical foundations of representation theorems for (unit-root) VAR models constitute different facets of the same economic variables of interest, derived from a dynamic model analysis, and the main purpose of the present monograph. The brief contents of the (three) enclosed chapters is as follows: Chapter 1 (The Algebraic Framework of Unit-Root Econometrics) is designed to provide the reader with a self-contained treatment of matrix theory aimed to prepare the way to a rigorous derivation of representation theorems. Several results are enclosed on generalized inverses, orthogonal complements, partitioned inversion rules, matrix polynomial inversion via Laurent expansions in matrix form, the notion of Schur complement, and a novel partitioned inversion formula. Chapter 2 (The Statistical Setting) introduces the basic notions regarding the multivariate stochastic processes. In particular, the reader will find definitions of stationarity and integration, the principle of stationary processes is introduced, the connection between integrated and cointegrated processes as well as the role of cointegration are pointed out. Chapter 3 (Econometric Dynamic Models: from Classical Econometrics to Time Series Econometrics) outlines the relationship between structural and autoregressive models, examines the VAR specification and resulting processes, and focuses, as central topic, on representation theorems relying on time series econometrics approach.