init_ssvs {bvhar}

Initial Parameters of Stochastic Search Variable Selection (SSVS) Model

Description

Set initial parameters before starting Gibbs sampler for SSVS.

Usage

init_ssvs(
  init_coef,
  init_coef_dummy,
  init_chol,
  init_chol_dummy,
  type = c("user", "auto")
)

## S3 method for class 'ssvsinit'
print(x, digits = max(3L, getOption("digits") - 3L), ...)

## S3 method for class 'ssvsinit'
knit_print(x, ...)

Arguments

init_coef

Initial coefficient matrix. Initialize with an array or list for multiple chains.

init_coef_dummy

Initial indicator matrix (1-0) corresponding to each component of coefficient. Initialize with an array or list for multiple chains.

init_chol

Initial cholesky factor (upper triangular). Initialize with an array or list for multiple chains.

init_chol_dummy

Initial indicator matrix (1-0) corresponding to each component of cholesky factor. Initialize with an array or list for multiple chains.

type

Type to choose initial values. One of "user" (User-given) and "auto" (OLS for coefficients and 1 for dummy).

x

ssvsinit

digits

digit option to print

...

not used

Details

Set SSVS initialization for the VAR model.

• init_coef: (kp + 1) x m A coefficient matrix.

• init_coef_dummy: kp x m \Gamma dummy matrix to restrict the coefficients.

• init_chol: k x k \Psi upper triangular cholesky factor, which \Psi \Psi^\intercal = \Sigma_e^{-1}.

• init_chol_dummy: k x k \Omega upper triangular dummy matrix to restrict the cholesky factor.

Denote that init_chol and init_chol_dummy should be upper_triangular or the function gives error.

For parallel chain initialization, assign three-dimensional array or three-length list.

Value

ssvsinit object

References

George, E. I., & McCulloch, R. E. (1993). Variable Selection via Gibbs Sampling. Journal of the American Statistical Association, 88(423), 881–889.

George, E. I., Sun, D., & Ni, S. (2008). Bayesian stochastic search for VAR model restrictions. Journal of Econometrics, 142(1), 553–580.

Koop, G., & Korobilis, D. (2009). Bayesian Multivariate Time Series Methods for Empirical Macroeconomics. Foundations and Trends® in Econometrics, 3(4), 267–358.