Flash estimation of GDP by production approach has been stated in Statistics Lithuania. Econometric methods are used for estimation of GDP components. The main problem is that time series are quite short and GDP is affected by results of work of single enterprises. That’s why it’s very important to know if the parameters of the models are stable or not.
Main purpose of this paper is to evaluate stability of the GDP components models. This is done by generation of random series (with Gaussian and Student distributions) as a new residuals and recalculation of parameters with new dependent variables, which are equal to the already estimated model plus generated residuals.
Main results of this work are quite satisfactory – new estimators are very close to first ones (most cases the error is less than 1 per cent). Exceptions are only the time series where residuals had outliers. Suggestion is to evaluate outliers prior to parameters estimation.
This work is licensed under a Creative Commons Attribution 4.0 International License.