Threshold Autoregression Monte Carlo Simulation in R or Eviews

Hi there,
I´m looking for support concerning a Threshold Autoregression Monte Carlo Simulation in R or Eviews.

First, let´s take an easy time series with
yt=ϕy(t-1)+εt
The time series starts at 100, that is: y0 = 100
εt is a normally distributed term with µ = 0 and σ = adjustable, let´s say it´s 0.025

Second, an AR(1) threshold regression is computed with 2 fix thresholds. The threshold varying regressor is y(t-1).

As a result, we get for example something like:
threshold1 = -0.3 and
threshold2 = 0.4

That means
ϕ1 = ? when y(t-1) < -0.3
ϕ2 = ? when -0.3 < y(t-1) < 0.4
ϕ3 = ? when y(t-1) > 0.4

A next step is to randomly simulate
yt=ϕy(t-1)+εt
27 times with different, adjustable values for σ.

After that, I´d like to compute
In how many out of 27 cases is ϕ1 < 1 and, simultaneously, is ϕ3 < 1

As a result, we get 1 statement like "in 4 cases ϕ1 and ϕ3 is <1". This test statistic (which is 4 in this case) is labeled θ

This process has to be repeated 10.000 times to get a distribution for θ.

I´m not sure if Eviews can handle something like that, but in R it should be possible.

Please let me know if you are able to handle this.