N/A
Re: Data Generation
I assume that you mean e(t-1), not e(t), in the evolution of h. This generates 500 data points. A certain number of early ones (maybe 100) should be discarded as burn-in draws.
Code: Select all
all 500
*
* Generate the fundamental shocks
*
set z = %ran(1.0)
set v = %ran(1.0)
*
* Generate the exogenous x
*
set(first=0.0) x = 1.0+.6*x{1}+z
*
* Start with e and h as their unconditional means
*
set e = 0.0
set h = 1.0/(1-.4-.3)
*
* Generate e and h together from entries 2 on
*
set e 2 * = h=1.0+.4*e{1}^2+.3*h{1},v*sqrt(h)
*
* Generate y given x, h and e series
*
set y = 2.0+3.0*x+.8*sqrt(h)+e