Using Structural and Nonstructural Shocks in the Estimation of DSGE Models
-
Series
-
Speaker(s)Fabio Canova (BI Norwegian Business School, Norway)
-
FieldEconometrics
-
LocationUniversity of Amsterdam, Roeterseilandcampus, room E0.22
Amsterdam -
Date and time
February 24, 2023
12:30 - 13:30
Dynamic Stochastic General Equilibrium (DSGE) models are typically singular. Thus, for likelihood-based structural parameters estimation, one needs to select the variables to use or add artificial disturbances to remove the singularity. Alternatively, one could use a composite likelihood of non -singular models. I compare various estimation approaches, provide conditions for the additional shocks to remove singularity, study shock identification and parameter estimates distortions; and reexamine the sources of macroeconomic fluctuations. Adding shocks may be counterproductive. A composite likelihood approach minimizes the distortions. The importance of monetary shocks for inflation is likely to be underestimated; TFP shocks are driving output fluctuations.