MACROECONOMETRICS

1
MACROECONOMETRICS

• LAB 5 SVARs

2
ROADMAP

• Some things by definition not linked with others
  ...
• Estimating SVARs
• Estimation
• Postestimation (diagnostics)
• All about STATA ?

3
Dataset (the same as previously)

• Timing
• Quarterly data, Jan. 1995 Dec. 2004
• Data
• GDP
• Consumption
• GDP deflator
• M3
• Quarterly dummies

4
STATA (1)

• set memory 99m
• set matsize 800
• Open downloaded file
• Real and logs are already generated ?

5
STATA (2)

• Defining SVARs
• Defining matrices
• aeq(matrix_aeq) - set of equality constraints.
• must be square with dimension equal to the number
  of equations in the underlying VAR.
• the elements must be missing or real numbers.
• Defining constraints
• acns(matrix_acns) - set of exclusion or
  cross-parameter equality constraints on A.
• must be square with dimension equal to the number
  of equations in the underlying VAR.
• the elements must be missing, 0, or a positive
  integer.
• Beq and bcns the same way!

6
STATA (3)

• SVARs
• matrix A (1, 0 ,0\ ., 1, 0\ ., ., 1)
• svar lrgdp lrcons lm3, aeq(A)
• And this is it !

7
STATA (4)

• How do we know if this SVAR is a good one?
• You do a VAR before
• You run complete diagnostics on this VAR
• varsoc (for no. of lags)
• varstable (for eigen values)
• varlmar (for autocorrelation)
• If its OK., AND you have theory behind your SVAR
  it should be OK...

8
STATA (5)

• Forecasting
• Computing
• svarfcast compute, step(8) dynamic(q(2005.1))
• svarfcast compute f_, step(8) replace
  dynamic(q(2005.1)) bs
• Graphing
• svarfcast graph lrconsumption ldef_CPI lm3 lrpkb