Simulate VAR data with heteroskedastic shocks and given parameters

Simulates data from a VAR(P) model with two types of structural shocks: regular shocks (always present) and event shocks (occurring every Nevn periods). The shock distribution can be normal, Student-t, or GARCH(1,1).

Usage

simulatedata(Phi, SigE, PsiE, PsiR, Nobs, Nbin, N, R, E, Nevn, P, eDist, seed)

Arguments

Phi

Array of VAR coefficient matrices, dimension N x N x P.

SigE

Variance of the event shocks. This applies to all E shocks.

PsiE

Impact matrix for event shocks, dimension N x E.

PsiR

Impact matrix for regular shocks, dimension N x R.

Nobs

Number of observations to retain after burn-in. The returned data have exactly Nobs rows.

Nbin

Number of burn-in observations. These are simulated to initialise the VAR but are discarded before returning.

N

Number of variables in the VAR.

R

Number of regular shocks.

E

Number of event shocks. For GARCH shock distributions (eDist = c(alpha, beta)), only E = 1 is supported.

Nevn

Event frequency: an event shock occurs every Nevn periods. Set to 0 to suppress event shocks entirely (no heteroskedasticity).

P

VAR lag order.

eDist

Shock distribution. Use 0 for standard normal; a positive integer for Student-t with that many degrees of freedom; or a numeric vector c(alpha, beta) for GARCH(1,1) with ARCH effect alpha and persistence beta.

seed

Integer seed passed to set.seed() for reproducibility. Use NA to skip seeding.

Value

A named list with components:

y

Simulated VAR data, dimension Nobs x N (burn-in discarded).

IndE

Event indicator matrix, dimension Nobs x 1.

eR

Simulated regular shocks, dimension Nobs x R.

eE

Simulated event shocks, dimension Nobs x E.

e

Composite structural shocks, dimension Nobs x N.

Phi

VAR coefficient array (returned unchanged).

PsiE

Event shock impact matrix (returned unchanged).

PsiR

Regular shock impact matrix (returned unchanged).

SigE

Event shock variance (returned unchanged).