Housework 3 (B)VAR model
1. [3p.] Come up with an interesting economic system that can be written down as a SVAR model.
Explain the theoretical framework, propose the SVAR representation (short-run vs. long-run identification)
2. [2p.] Gather and present the data.
3. [2p.] Estimate the VAR model (select lags, check for autocorrelation) 4. [1p.] Interpret identification matrices of SVAR
5. [3p.] Plot and interpret IRFs, FEVD, historical decomposition (levels vs. growth rates)
6. [2p.] Estimate a BVAR model (set mean prior for own lags to 0, 0.9 or 1 depending on persistence). Compare posterior mean and LS estimates for B parameters
7. [3p.] Calculate Bayesian IRFs
8. [3p.] Calculate forecast from VAR/BVAR for the next two years. Draw the forecast on the same graph and explain the differences.
9. [1p.] Make a general discussion about the usefulness of VAR models for the addressed topic
or forecasting competition
1. [3p.] Come up with an interesting forecasting topic (related to VAR/BVAR).
Present the results already available in the scientific literature (refer at least to three articles) 2. [2p.] Present the models participating in your forecasting competition
(VAR/BVAR should be one of the participating methods) 3. [2p.] Gather and present the data.
4. [3p.] Make a graph of sequential forecasts / scatter plots of realizations on forecasts 5. [1p.] Present the results for mean error
6. [3p.] Present the results for RMSE and DM test
7. [3p.] Present the results for log-predictive scores and GA test 8. [2p.] Present the results for PIT
9. [1p.] Make a general discussion about the usefulness of VAR models in forecasting