Use of models at central banks: Comments by

Use of models at central banks:
Comments
Gunnar Bårdsen
Department of Economics
NTNU, Trondheim
Monetary Policy Conference at Norges Bank
7 April 2005
Gunnar Bårdsen (NTNU)
Models: Comments
1/8
Possibilities
Demand & supply
At the model shop
What would you like?
Communication
Reality check
Forecasting
Policy analysis
Gunnar Bårdsen (NTNU)
In stock
Calibrated theory
VAR (Structural &
Bayesian)
Econometric model
DSGE
Models: Comments
2/8
Possibilities
What to take?
What to take?
Calibrated theory
I
non-realistic
VAR & econometric
model
I
Winnie the Pooh solution:
Yes, please.
non-constant
DSGE
I
non-robust
Gunnar Bårdsen (NTNU)
Models: Comments
3/8
The devil’s advocate
The devil’s dictionary
The devil’s dictionary
Apologies to Ambrose Bierce
Aggregate dynamic econometric models
I
Estimates of "something"
Structural VARs
I
How to estimate models inefficiently
Bayesian estimation
I
Calibration called estimation
DSGE & Microfoundations
I
Sophisticated naivety
Gunnar Bårdsen (NTNU)
Models: Comments
4/8
Core considerations
What to choose
What to choose
"Core" or "suite"
Why?
Why not core?
What suite?
Suite
Competition & diversification
Models come...models go...
"Story"-model
Data-based forecasting model
I
Must be transparent
Laboratory
Gunnar Bårdsen (NTNU)
Models: Comments
5/8
Core considerations
How much reality?
How much reality?
The Pagan frontier...
Theory
DSGE
VAR
Data
Gunnar Bårdsen (NTNU)
Models: Comments
6/8
Core considerations
How much reality?
How much reality?
The Pagan frontier...
Rejected
theory
Theory
Theory
DSGE
VAR
Data
Gunnar Bårdsen (NTNU)
Models: Comments
6/8
Core considerations
How much reality?
How much reality?
The Pagan frontier...
Theory
DSGE
VAR
Data
Gunnar Bårdsen (NTNU)
Models: Comments
6/8
Core considerations
Microfoundations...
Microfoundations...
The times they are achanging
Behavioural economics
hyperbolic discounting: time inconsistency
learning: expectations change
herd behaviour: asset bubbles
fair treatment: backward-looking behaviour (Driscoll &
Holden)
But see Backus, Routledge & Zin: "Exotic preferences for
macroeconomists"
Gunnar Bårdsen (NTNU)
Models: Comments
7/8
Core considerations
If core
If core...
Danger: vested interests
I
Any new model is a success by default
Economies are highly non-linear & non-constant
Go for robustness & pragmatism:
Build in blocks
Pick "best" theory-consistent model for each block
If any common framework: steady-state medium term
growth
Bårdsen, Hurn & Lindsay: approximations imply EqCM
Cointegration provides the steady-state.
Gunnar Bårdsen (NTNU)
Models: Comments
8/8