MacroEconometrics

Roadmap

Basics

• VAR model

  • Simulation
  • Bayesian
    • Minnesota
    • SSVS
    • Indpendent Normal-Wishart
  • Frequentist
    • LS
    • Bootstrap CI
  • Forecasting

• SVAR model

  • Simulation
  • Estimation --> Direct
  • Identification
    • Cholesky
    • Short-run restrictions
    • Long-run restrictions
    • Heteroskedastic
    • Sign
    • Proxy VAR

• Impulse Response Functions

  • Reduced Form
    • From VAR
  • Structural
    • From SVAR
    • Identification from VAR
    • Instrumental Variables

• Local Projections

• Visualisation

• Data Wrangling

• Replicability

• FEVD

  • SVAR

• Historical Decomposition

  • SVAR

• Forecast Scenarios

  • SVAR

• Policy Counterfactuals

Basic Structure

• Every macroeconometric model should just contain the most basic structure
  • It should contain only those elements that one would also have to write down, abstracting away from how it is estimated.
  • Parts that need to be estimated should be a subtype of Estimated which should either be Frequentist or Bayesian
  • If a part is fixed (because it was used in a simulation), then it should be of type Fixed. These parts are as if we would completely know them
• IRFs, Forecasts, FEVD, ... should all use the Estimated type of parts of their results depend on estimations.