mat_forc: mat_forc function...

In APRScenario: Structural Scenario Analysis for Bayesian Structural Vector Autoregression Models

mat_forc function ############################################################################## NB: HERE WE USE Antolin-Diaz et al notation # B is reduced form; # A is structural; # d is intercepts # M is reduced so that E(uu')=Sigma=(A_0A_0')^(-1) and M_0=A_0^(-1)*Q # Note that the code returns conflicting notation: # B=>A_0^(-1)*Q and # A=>B # ##############################################################################

Description

mat_forc function ############################################################################## NB: HERE WE USE Antolin-Diaz et al notation # B is reduced form; # A is structural; # d is intercepts # M is reduced so that E(uu')=Sigma=(A_0A_0')^(-1) and M_0=A_0^(-1)*Q # Note that the code returns conflicting notation: # B=>A_0^(-1)*Q and # A=>B # ##############################################################################

Usage

mat_forc(
  h = 1,
  n_draws,
  n_var,
  n_p,
  data_ = NULL,
  matrices = NULL,
  max_cores = 1
)

Arguments

h	(integer) forecast horison
n_draws	(integer) Number of draws
n_var	(integer) Number of variables
n_p	(integer) Number of lags
data_	(matrix optional) The data, stacking Y over X (data and laggs) – columns are observations (default taken from matrices$Z) NB: this is not necessarily the same as the data used to estimate the model If run counterfactuals in previoius historical period (ie not forecast) must pass the data up to previous period relative to counterfactual
matrices	Optional matrices object from gen_mats() (default taken from calling environment)
max_cores	maximum number of cores to use for parallel processing (default: 1 for Windows compatibility)

Value

the big_b and big_M matrices of mean and IRF

Examples

library(APRScenario)
data(NKdata)

# Minimal example with a toy specification
spec <- bsvarSIGNs::specify_bsvarSIGN$new(as.matrix(NKdata[,2:4]), p = 1)
est <- bsvars::estimate(spec, S = 10)  # Use small S for fast test
matrices<-gen_mats(posterior = est, specification = spec)

# Example usage for matrix forecasting
result <- mat_forc(h = 4, n_draws = 10, n_var = 3, n_p = 1,
                   matrices = matrices)

APRScenario documentation built on Dec. 22, 2025, 1:06 a.m.