Fix Required: Cant perform this Rolling Regression, MLE, folows ARMA(1,1), asreg

[SM Muzammil]

HI all,

I am trying to find the residuals of the following ARMA (1,1): ILL_i,t =α_0,i +α_1,i ILL_i,t−1+α_2,iε_i,t−1+ε_i,t,

ILL is an illiquidity measure.

The parameters required estimated via MLE using a 12-month rolling sample.

My main motive here is to get the residuals.

Been trying this code but this isn't helping;

local rolling_window 24
forvalues i = `rolling_window' / `=_N' {
local start_obs = `=_N' - `i' + 1
local end_obs = `=_N' - `i' + `rolling_window'
arima ILL[`start_obs'/`end_obs'], arima(1,0,1) noconstant
predict fitted_values[`start_obs'/`end_obs'], xb
gen residuals[`start_obs'/`end_obs'] = ILL[`start_obs'/`end_obs'] - fitted_values[`start_obs'/`end_obs']
}

The above isn't working

have also tries asreg:
bys ID : asreg ILL L.ILL L.Residuals, wind( mofd 12)

but it is returning this error "factor variables and time-series operators not allowed"

Please can anyone help me in this.

Thanks

[Leonardo Guizzetti]

Not an economist or econometrician so I can’t help further them to point out you’re rolling regression loop is not remotely valid Stata syntax. I don’t know where you got it from but it looks like the kind of crud from a GPT type model.

asreg is from SSC as you are asked to explain in the FAQ. I also don’t use this so I can’t comment on whether this is how it should work.