Autoregressive Distributed Lag Models with MA component

Hello,

As the title suggests I try to estimate a time series model that incorporates several lag for different series, but with a MA component. Since I want to incorporate my term delay error I can not estimate my model by OLS so I have to use a non-linear method as the maximum likelihood. Someone could help me because I know what is the function that I have to minimize, but I can not implement it. Dufour explains how to do well with an arma(p, q) model ( http://www2.cirano.qc.ca/~dufourj/Web_Site/ResE/Dufour_2008_C_TS_ARIMA_Estimation.pdf page 3) which is similar in my case but with other time series.

Thank you

[Amor-Aniss Benmoussa]

This what I try.

lag = 1;

T = length(DataY);

e = 0;

logL = 0;

for t = lag+1:T

fun = @(beta) (DataY(t,1) - beta(1) - beta(2)*DataY(t-1,1) - beta(3)*e);

e = @(beta) fun;

logL = @(beta) (fun + logL);

end

beta0 = [.5,.5,.5];

min = lsqnonlin(logL,beta0)

[Brendan Hamm]

Do you have the econometrics toolbox? This would likely make this easier.

These lines make no sense:

e = @(beta) fun;

logL = @(beta) (fun + logL);

What are you trying to do with them?

[Amor-Aniss Benmoussa]

I want to estimate the parameters of a arima(1,0,1) model using non linear least-squares solver. I do not want to use the Matlab function already done, because I have to understand how to implement the MA component to add it in an another econometric model. And yes I have all the necessary toolbox.

Thank you