Structural vector autoregressive analysis / by Lutz Kilian and Helmut Lütkepohl.

Includes bibliographical reference and index

Table of contents Cover
Half title
Series page
Title
Copyright
Contents
Preface
1 Introduction
1.1 Overview
1.2 Outline of the Book
2 Vector Autoregressive Models
2.1 Stationary and Trending Processes
2.2 Linear VAR Processes
2.2.1 The Basic Model
2.2.2 The Moving Average Representation
2.2.3 VAR Models as an Approximation to VARMA Processes
2.2.4 Marginal Processes, Measurement Errors, Aggregation, Variable Transformations
2.3 Estimation of VAR Models
2.3.1 Least-Squares Estimation
2.3.2 Restricted Generalized Least Squares
2.3.3 Bias-Corrected LS
2.3.4 Maximum Likelihood Estimation
2.3.5 VAR Processes in Levels with Integrated Variables
2.3.6 Sieve Autoregressions
2.4 Prediction
2.4.1 Predicting from Known VAR Processes
2.4.2 Predicting from Estimated VAR Processes
2.5 Granger Causality Analysis
2.6 Lag-Order Selection Procedures
2.6.1 Top-Down Sequential Testing
2.6.2 Bottom-Up Sequential Testing
2.6.3 Information Criteria
2.6.4 Recursive Mean-Squared Prediction Error Rankings
2.6.5 The Relative Merits of Alternative Lag-Order Selection Tools
2.7 Model Diagnostics
2.7.1 Tests for Autocorrelation in the Innovations
2.7.2 Tests for Nonnormality
2.7.3 Residual ARCH Tests
2.7.4 Time Invariance
2.8 Subset VAR Models, AVAR Models, and VARX Models
2.8.1 Subset VAR Models
2.8.2 Asymmetric VAR Models
2.8.3 VARX Models
3 Vector Error Correction Models
3.1 Cointegrated Variables and Vector Error Correction Models
3.1.1 Common Trends and Cointegration
3.1.2 Deterministic Terms in Cointegrated Processes
3.2 Estimation of VARs with Integrated Variables
3.2.1 The VAR(1) Case
3.2.2 Estimation of VECMs
3.2.3 Estimation of Levels VAR Models with Integrated Variables. 6.2.5 Summary of Potential Problems in Approximating DSGE Models with VAR Models
6.3 DSGE Models as an Alternative to VAR Models?
6.3.1 Calibrated DSGE Models
6.3.2 Estimated DSGE Models
6.3.3 Calibration versus Bayesian Estimation
6.3.4 Are Structural VAR Models Less Credible than DSGE Models?
6.3.5 Are DSGE Models More Accurate than VAR Models?
6.3.6 Policy Analysis in DSGE Models and SVAR Models
6.4 An Overview of Alternative Structural Macroeconometric Models
6.4.1 Combining DSEMs and SVAR Models
6.4.2 Combining DSGE and SVAR Models
7 A Historical Perspective on Causal Inference in Macroeconometrics
7.1 A Motivating Example
7.2 Granger Causality Tests for Covariance Stationary VAR Models
7.3 Granger Causality, Predeterminedness, and Exogeneity
7.3.1 Basic Concepts
7.3.2 Granger Causality and Forward-Looking Behavior
7.3.3 Strict Exogeneity in Modern Macroeconomic Models
7.4 The Demise of Granger Causality Tests in Macroeconomics
7.5 Responses to Unanticipated Changes in Money Growth
7.5.1 The Narrative Approach
7.5.2 Exogenous Shocks Derived from Data-Based Counterfactuals
7.5.3 News Shocks
7.5.4 Shocks to Financial Market Expectations
7.5.5 Summary
7.6 Structural VAR Shocks
7.6.1 The Identification Problem
7.6.2 The Relationship between Structural VAR Shocks and Direct Shock Measures
7.6.3 Causality in Structural VAR Models
8 Identification by Short-Run Restrictions
8.1 Introduction
8.2 Recursively Identified Models
8.3 Sources of Identifying Restrictions
8.4 Examples of Recursively Identified Models
8.4.1 A Simple Macroeconomic Model
8.4.2 A Model of the Global Market for Crude Oil
8.4.3 Oil Price Shocks and Stock Returns
8.4.4 Models of the Transmission of Energy Price Shocks. 8.4.5 Semistructural Models of Monetary Policy
8.4.6 The Permanent Income Model of Consumption
8.5 Examples of Nonrecursively Identified Models
8.5.1 Fiscal Policy Shocks
8.5.2 An Alternative Simple Macroeconomic Model
8.5.3 Discussion
8.5.4 The Graph-Theoretic Approach
8.6 Summary
9 Estimation Subject to Short-Run Restrictions
9.1 Model Setup
9.2 Method-of-Moments Estimation
9.2.1 Recursively Identified Models
9.2.2 Nonrecursively Identified Models
9.2.3 GMM Estimation of Overidentified Models
9.3 Instrumental Variable Estimation
9.4 Full Information Maximum Likelihood Estimation
9.5 Bayesian Estimation
9.6 Summary
10 Identification by Long-Run Restrictions
10.1 The Traditional Framework for Imposing Long-Run Restrictions
10.2 A General Framework for Imposing Long-Run Restrictions
10.2.1 The Long-Run Multiplier Matrix
10.2.2 Identification of Structural Shocks
10.3 Examples of Long-Run Restrictions
10.3.1 A Real Business Cycle Model with and without Nominal Variables
10.3.2 A Model of Neutral and Investment-Specific Technology Shocks
10.3.3 A Model of Real and Nominal Exchange Rate Shocks
10.3.4 A Model of Expectations about Future Productivity
10.4 Examples of Models Combining Long-Run and Short-Run Zero Restrictions
10.4.1 The IS-LM Model Revisited
10.4.2 A Model of the Neoclassical Synthesis
10.4.3 A U.S. Macroeconomic Model
10.5 Limitations of Long-Run Restrictions
10.5.1 Long-Run Restrictions Require Exact Unit Roots
10.5.2 Sensitivity to Omitted Variables
10.5.3 Lack of Robustness at Lower Data Frequencies
10.5.4 Nonuniqueness Problems without Additional Sign Restrictions
10.5.5 Sensitivity to Data Transformations
11 Estimation Subject to Long-Run Restrictions
11.1 Model Setup. 11.2 Models Subject to Long-Run Restrictions Only
11.2.1 Method-of-Moments Estimation
11.2.2 Full Information Maximum Likelihood Estimation
11.2.3 Instrumental Variable Estimation
11.3 Models Subject to Long-Run and Short-Run Restrictons
11.3.1 Estimating the Model in VAR Representation
11.3.2 Estimating the Model in VECM Representation
11.4 Practical Limitations of Long-Run Restrictions
11.4.1 Estimators of the Long-Run Multiplier Matrix May Be Unreliable
11.4.2 Lack of Power
11.4.3 Near-Observational Equivalence of Shocks with Permanent Effects and Shocks with Persistent Effects
11.4.4 Weak Instrument Problems
11.5 Can Structural VAR Models Recover Responses in DSGE Models?
11.5.1 The Origin of This Controversy
11.5.2 The Position of Chari et al. (2008)
11.5.3 The Position of Christiano et al. (2006)
11.5.4 Understanding the Simulation Evidence
11.5.5 Summary
12 Inference in Models Identified by Short-Run or Long-Run Restrictions
12.1 Delta Method Intervals for Structural Impulse Responses
12.1.1 Finite-Order VAR Models
12.1.2 Infinite-Order VAR Models
12.1.3 Discussion
12.1.4 Extensions to Other Statistics
12.1.5 On the Choice of the Significance Level
12.2 Bootstrap Intervals for Structural Impulse Responses
12.2.1 The Standard Residual-Based Recursive-Design Bootstrap
12.2.2 The Standard Residual-Based Fixed-Design Bootstrap
12.2.3 The Residual-Based Wild Bootstrap
12.2.4 Bootstrapping Tuples of Regressands and Regressors
12.2.5 Block Bootstrap Methods
12.2.6 Alternative Bootstrap Confidence Intervals
12.3 Bootstrap Intervals Based on Bias-Adjusted Estimators
12.4 Potential Pitfalls in Impulse Response Inference
12.5 Finite-Sample Properties of Bootstrap Confidence Intervals

Structural vector autoregressive (VAR) models are important tools for empirical work in macroeconomics, finance, and related fields. This book not only reviews the many alternative structural VAR approaches discussed in the literature, but also highlights their pros and cons in practice. It provides guidance to empirical researchers as to the most appropriate modeling choices, methods of estimating, and evaluating structural VAR models. The book traces the evolution of the structural VAR methodology and contrasts it with other common methodologies, including dynamic stochastic general equilibrium (DSGE) models. It is intended as a bridge between the often quite technical econometric literature on structural VAR modeling and the needs of empirical researchers. The focus is not on providing the most rigorous theoretical arguments, but on enhancing the reader's understanding of the methods in question and their assumptions. Empirical examples are provided for illustration

Computer Science & Engineering Computer Science & Engineering

Golam Newaz