# Length of time series and likelihood estimation

In ARMA (with normal errors) model estimation, are there any empirical studies or tests to judge the minimum number of observations (length) of the time series that are required such that OLS is an acceptable estimation of the model (in contrast to an exact likelihood)?

• You cannot estimate MA part coefficients with OLS, and if there is no MA part, OLS estimate is MLE. So please clarify the question. – mpiktas Apr 28 '13 at 6:40
• @mpiktas OLS estimates conditional likelihood, and it is not accurate for time series with rather limited sample size, thats where exact likelihood should be used to represent MLE instead of using OLS. – user55647 Apr 28 '13 at 7:23