garch_modelspec function

GARCH Model Specification

Specifies a GARCH model prior to estimation.

garch_modelspec(
  y,
  model = "garch",
  constant = FALSE,
  order = c(1, 1),
  variance_targeting = FALSE,
  vreg = NULL,
  multiplicative = FALSE,
  init = c("unconditional", "sample", "backcast"),
  backcast_lambda = 0.7,
  sample_n = 10,
  distribution = "norm",
  ...
)

Arguments

• y: an xts vector.
• model: the type of GARCH model. Valid choices are garch for vanilla GARCH, gjr for asymmetric GARCH, egarch for exponential GARCH, aparch for asymmetric power ARCH, csGARCH for the component GARCH, igarch for the integrated GARCH.
• constant: whether to estimate a constant (mean) for y,
• order: the (p,q) GARCH order.
• variance_targeting: whether to use variance targeting rather than estimating the conditional variance intercept.
• vreg: an optional xts matrix of regressors in the conditional variance equation.
• multiplicative: whether to exponentiate the contribution of the regressors else will be additive. In the case of the egarch model, since this is already a multiplicative model, the regressors are additive irrespective of the choice made.
• init: the method to use to initialize the recursion of the conditional variance.
• backcast_lambda: the decay power for the exponential smoothing used when initializing the recursion using the backcast method.
• sample_n: the number of data points to use when initializing the recursion using the sample method.
• distribution: a valid distribution from the available re-parameterized distributions of the package.
• ...: not used.

Returns

An object of class tsgarch.spec .

Details

The specification object holds the information and data which is then passed to the maximum likelihood estimation routines.

Author(s)

Alexios Galanos