fsvsim generates simulated data from a factor SV model.
fsvsim(
n = 1000,
series = 10,
factors = 1,
facload = "dense",
idipara,
facpara,
heteroskedastic = rep(TRUE, series + factors),
df = Inf
)The value returned is a list object of class fsvsim holding
The simulated data, stored in a n times m matrix with
colnames 'Sim1', 'Sim2', etc.
The simulated factors, stored in a r times r matrix.
Factor loadings matrix.
Latent factor log-variances for times 1 to n.
Initial factor log-variances for time 0.
The parameters of the factor volatility processes.
Latent idiosyncratic log-variances for times 1 to n.
Initial idiosyncratic log-variances for time 0.
The parameters of the idiosyncratic volatility processes.
Length of the series to be generated.
Number of component series m.
Number of factors r.
Can either be a matrix of dimension m times r
or one of the keywords "dense" and "sparse". If "dense" is chosen,
a (rather) dense lower triangular factor loadings matrix is randomly
generated. If "sparse" is chosen, a (rather) sparse lower triangular
factor loadings matrix is randomly generated.
Optional matrix of idiosyncratic SV parameters
to be used for simulation. Must have exactly three columns containing
the values of mu, phi and sigma for each
of m series, respectively. If omitted, plausible values are
generated.
Optional matrix of idiosyncratic SV parameters
to be used for simulation. Must have exactly two columns containing
the values of phi and sigma for each of r factors,
respectively. If omitted, plausible values are generated.
Logical vector of length m+r. When
TRUE, time-varying volatilities are generated; when
FALSE, constant volatilities (equal to mu) are generated.
If not equal to Inf, the factors are misspecified (come from a t distribution instead of a Gaussian). Only used for testing.