European Centre for Advanced Studies in Economics and Statistics Solvay Brussels School of Economics and Management
Combining structural and reduced-form models for macroeconomic forecasting and policy analysis
Dissertation pr´esent´ee en vue de l’obtention du titre de Docteur en Sciences ´economiques et de gestion
Le 8 F´evrier 2011
par
Francesca Monti
sous la direction des ProfesseursDomenico Giannone etPhilippe Weil
Membres du jury:
BramDe Rock Universit´e Libre de Bruxelles MarcoDel Negro Federal Reserve Bank of New York MarjorieGassner Universit´e Libre de Bruxelles Domenico Giannone Universit´e Libre de Bruxelles RobertKollman Universit´e Libre de Bruxelles Lucrezia Reichlin London Business School David Veredas Universit´e Libre de Bruxelles PhilippeWeil Universit´e Libre de Bruxelles Rafael Wouters National Bank of Belgium
Contents
Introduction vii
1 Combining Judgment and Models 1
1.1 Introduction . . . 1
1.2 The Econometric Methodology . . . 4
1.2.1 The Framework . . . 4
1.2.2 Model of the Judgmental Forecasts . . . 5
1.2.3 Using the model to interpret judgemental forecasts . . . 10
1.3 An application . . . 11
1.4 Forecasting and Structural Analysis . . . 17
1.5 Conclusions . . . 26
2 Incorporating conjunctural analysis in structural models 27 2.1 Introduction . . . 27
2.2 The methodology . . . 29
2.3 Design of the Forecasting Exercise . . . 34
2.4 Empirical results . . . 39
2.4.1 Forecast Accuracy . . . 39
2.4.2 Structural analysis . . . 47
2.5 Conclusions . . . 51
3 Identifying misspecification in a data-rich environment 53 3.1 Introduction . . . 53
3.2 Misspecification and Granger causality . . . 55
3.2.1 Granger-causality . . . 55 v
vi CONTENTS
3.2.2 What is misspecification? . . . 57
3.3 Bivariate Granger causality tests . . . 61
3.4 Conclusions . . . 65
4 Further research: A multivariate framework for misspecification 67 4.1 Introduction . . . 67
4.2 The Methodology . . . 68
4.2.1 Generalized dummy observations (Sims, 2008) . . . 69
4.2.2 The DSGE-VARX . . . 71
4.3 The Application . . . 73
4.4 Conclusions . . . 83
General Conclusions 85
Appendix A 87
Appendix B 92
Appendix C 94
Bibliography 97