mcSimulations {sparsevar}

Monte Carlo simulations

Description

This function generates Monte Carlo simulations of sparse VAR and its estimation (at the moment only for VAR(1) processes).

Usage

mcSimulations(
  N,
  nobs = 250,
  nMC = 100,
  rho = 0.5,
  sparsity = 0.05,
  penalty = "ENET",
  covariance = "Toeplitz",
  method = "normal",
  modelSel = "cv",
  ...
)

Arguments

N

dimension of the multivariate time series.

nobs

number of observations to be generated.

nMC

number of Monte Carlo simulations.

rho

base value for the covariance.

sparsity

density of non zero entries of the VAR matrices.

penalty

penalty function to use for LS estimation. Possible values are "ENET", "SCAD" or "MCP".

covariance

type of covariance matrix to be used in the generation of the sparse VAR model.

method

which type of distribution to use in the generation of the entries of the matrices.

modelSel

select which model selection criteria to use ("cv" or "timeslice").

...

(TODO: complete)

Value

a nMcx5 matrix with the results of the Monte Carlo estimation