bayesVAR_TVP: Estimates TVP-VAR model using MCMC sampler.

In GediminasB/bayesVAR_TVP: Bayesian analysis of Vector Autoregressive Models

Description

Estimates Time-Varying Parameters VAR model using MCMC sampler.

Y_t = Z_t β_t + ε_t; ε_t ~ N(0, H)

β_t = β_{t-1} + u_t; u_t ~ N(0, Q)

Prior parameters are estimated using OLS method on training sample.

Usage

bayesVAR_TVP(Y, p = 1, nburn = 10000, nsim = 50000, tau = 40,
  beta.algorithm = "DK", reject.explosive = FALSE)

Arguments

Y	Matrix or timeseries object
p	Integer for the lag order (default is p=1)
nburn	Number of MCMC draws used to initialize the sampler (defaults to 10000)
nsim	Number of MCMC draws excluding burn-in (defaults to 50000)
tau	Length of training sample used for determining prior parameters via OLS
beta.algorithm	Algorithm for drawing time-varying VAR parameters. Either 'DK' for Durbin and Coopman 2002, or 'CC' for Carter and Cohn 1994. Defaults to 'DK'
reject.explosive	Should the explosive (not stable) draws of VAR parameters be rejected? Defaults to 'FALSE'. For rationale consult Cogley and Sargent 2005. May significantly increase computtion time!

Value

beta	Draws of time-varying parameters (beta_t). 4D array [t x M x M*p+1 x draw]
H	Draws of observation equation error term covariance matrix H. 3D array [M x M x draw]
Q	Draws of state equation error term covariance matrix Q. 3D array [M x M x draw]

References

• \insertRef

  KoopKorobilis2010bayesVAR

• \insertRef

  CC1994bayesVAR

• \insertRef

  DK2002bayesVAR

• \insertRef

  CogleySargent2005bayesVAR

GediminasB/bayesVAR_TVP documentation built on Nov. 18, 2019, 6:44 p.m.