Understanding Vector Autoregressions

The book offers a concise, accessible guide to Vector Autoregression (VAR) models. Starting with data preparation, the book demonstrates how VAR models are estimated, the criteria for selecting the number of lags and how the researcher should use model diagnostics. Expanding the analysis to Structural VARs (SVAR), the book then illustrates, with numerical examples, how their most popular outputs (impulse response functions, forecast error variance decomposition, historical decomposition) should be calculated and interpreted. Other topics such as potential data issues, confidence intervals, and forecasting are also explored, along with the most usual identification schemes. VAR model extensions such as Local Projections, Cointegration (Vector Error Correction), and Bayesian Models are also included, offering examples to that aim to assist practitioners. The book includes more advanced topics such as time-varying VARs. Throughout the book, emphasis is placed on the interpretation and understanding of the models, and even though mathematical formulae are presented, this is for the purpose of linking the analysis with the existing literature on the topic. The book is an ideal resource for researchers, students and professional economists alike who wish to better understand econometric forecasting concepts without the burden of complex mathematical derivations.