Estimate conjugate Normal-Inverse-Wishart bayesian VAR model
bvar_conjugate0(Y_in = NULL, Z_in = NULL, constant = TRUE, p = NULL, keep = 10000, verbose = FALSE, priors = list(), fast_forecast = FALSE, way_omega_post_root = c("cholesky", "svd"))
| Y_in | the matrix or data.frame with endogeneous VAR variables |
|---|---|
| Z_in | (NULL by default) the matrix or data.frame with exogeneous VAR variables |
| constant | (TRUE by default) whether we should include constant |
| p | (2 by default) the number of lags |
| keep | (10000 by default) the number of Gibbs sampling replications to keep Is ignored when the fast_forecast is TRUE. |
| verbose | (FALSE by default) |
| priors | the list containing at least Phi_prior [k x m], Omega_prior [k x k], S_prior [m x m], v_prior [1x1], it may also contain Y_dummy [T_dummy x m], X_dummy [T_dummy x k] where k = mp+d |
| fast_forecast | logical, FALSE by default. If TRUE then no simulations are done, only posterior hyperparameters are calculated. |
| way_omega_post_root | the way for (Omega_post)^1/2 calculation: 'svd' or 'cholesky' |
the list containing all results of bayesian VAR estimation
Estimate conjugate Normal-Inverse-Wishart bayesian VAR model
data(Yraw) priors <- Carriero_priors(Yraw, p = 4) model <- bvar_conjugate0(priors = priors, keep = 100)