Simulating ARMA(2,1) process

[Kwasi Tabiri]

I am trying to simulate an ARMA(2,1) process using the following code:

clear
set obs 100
gen t=_n
tsset t


gen et = rnormal()   // generating epsilon_t
gen ut = et in 1

replace ut = 0.8*L1.ut - 0.6*L2.ut + et + 0.5*L1.et if t>1

For some reason, the rest of the observations for ut remain missing values after I run the last line of the code. I can't figure out exactly what I am doing wrong. I would really appreciate some help.

Thanks!

[Andrew Musau]

The second lag of "ut" is missing as you initially have

gen ut = et in 1

[.] So subtracting a missing value will result in missing values. A possible modification could be:

clear
set obs 100
gen t=_n
tsset t


gen et = rnormal()   // generating epsilon_t
gen ut = et if t<=2

replace ut = 0.8*L1.ut - 0.6*L2.ut + et + 0.5*L1.et if t>2