Identify the main shock by targetting the forecast error variance contribution in the frequency domain.

Usage

bootstrap(
  var,
  id_method = NULL,
  nboot = 500,
  n_ahead = 20,
  design = "recursive",
  method = "resample",
  wild_distr = "gaussian",
  bias_adjust = FALSE,
  lower_pctl = 0.16,
  upper_pctl = 0.84,
  ...
)

Arguments

var: VAR to bootstrap
id_method: function that identifies the Structural VAR
nboot: number of bootstrap iteratons
n_ahead: length of IRF horizon
design: "recursive" or "fixed". Controls how samples of the data are constructed.
method: "resample" or "wild"
wild_distr: distribution used for "wild" boostrap method. Either "gaussian", "rademacher", or "mammen".
bias_adjust: Boolean. Set to TRUE to calculate the bias and run the bootstrap again.
lower_pctl: lower percentile of bootstraps returned in summary. Defaults to 0.16 (for 68% CI).
upper_pctl: upper percentile of bootstraps returned in summary. Defaults to 0.84 (for 68% CI).
...: Additional arguments passed to the function passed to id_method

Value

list of bootstrapped VARs and IRFs

Examples

x <- svars::USA
v <- vars::VAR(x, p = 2)
mvar <- id_fevdtd(v, "pi", 4:10)
bmvar <- bootstrap(mvar, id_fevdtd, target = "pi", horizon = 4:10)
#> [ -------------------- ]
[ =------------------- ]
[ ==------------------ ]
[ ===----------------- ]
[ ====---------------- ]
[ =====--------------- ]
[ ======-------------- ]
[ =======------------- ]
[ ========------------ ]
[ =========----------- ]
[ ==========---------- ]
[ ===========--------- ]
[ ============-------- ]
[ =============------- ]
[ ==============------ ]
[ ===============----- ]
[ ================---- ]
[ =================--- ]
[ ==================-- ]
[ ===================- ]
[ ==================== ]