Description Usage Arguments Details References See Also Examples
Fit the parameter lambda of the Bayesian VAR. This parameter controls the importance given to the priors. If lambda=0 the model is the same as the OLS case. For bigger values of lambda more importance is given to the priors and less importance to the data.
1 2 3 4 5 6 7 8 9 10 11 12 |
Y |
Time-series matrix or data.frame with the VAR endogenous variables. |
variables |
Either a numeric vector indication the position of the variables to be included in the small model or a characted vector with the variable names. |
lambdaseq |
Sequence of lambdas to be tested. |
p |
Lag order (default = 1). |
p.reduced |
Lag order of the small model. |
delta |
Numeric vector indicating the prior for the autorregressive coefficients (default = 0 for all variables). If the prior is the same for all variables the user may supply a single number. Otherwise the vector must have one element for each variable. |
xreg |
Exogenous controls. |
scale |
If TRUE the variables are centered with variance equal 1 (default is TRUE). |
ps |
If TRUE the priors on the sum of the coefficients will be included. |
tau |
Controls the shrinkage in the priors on the sum of the coefficients. |
The choice of lambda is arbitrary. However, Banbura et al. (2010) uses the fit of a smaller model with just a few variables as a target to the Bayesian VAR. In other words, this function chooses the lambda that matches the fit of a smaller model chosen by the user on the chosen Bayesian VAR. If lambda = 0 the model ignores the data and the posterior equal the prior. For bigger lambda the model converges to the OLS estimates.
Banbura, M., Giannone, D., & Reichlin, L. (2010). Large Bayesian vector autoregressions. Journal of Applied Econometrics, 25, 71–92.
Garcia, Medeiros and Vasconcelos (2017).
1 2 3 4 5 6 7 8 |
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.