Econometric analysis of the general linear model with Gretl

In this exercise, using OLS, we estimate a model that seeks to explain consumption in terms of GDP. After entering the data, firstly, we represent the scatter graph of consumption against GDP (which also gives us the models estimate in the top section). We note that the linear relationship is suitable for the model considered. Following this graphical representation, we estimate the model using OLS, yielding the following results:

both coefficients being significant and with an R square of 0'9923. Therefore, it seems that this is a good fit.

However, we consider the option of studying the existence of autocorrelation in the model. To this end, we see on the results screen that there is also the Durbin-Watson statistic test (3'05). In addition, this also gives us the number of observations considered (6) and it is clear that the number of excluded regressors is a constant. Therefore, with this information, using the Durbin-Watson tables, we obtain the limits: du = 1'4002 and dl = 0'6102. Then, we find that the Durbin-Watson contrast is inconclusive. Then we turn to graphical methods. So in the screen where we estimate the model, we can represent the temporal graph of the residuals, observing an alternation in the sign of the same. This brings to mind the possible existence of negative autocorrelation. Another possible representation is the scatter graph of residuals compared to any of its delays. To that end, we keep the residuals, obtain the delayed series and represent them, seeing a downward trend which again made me consider the existence of negative autocorrelation.