When forecasts are obtained by fitting a model and computing minimum mean-square-error forecasts from the model, the differences between the observed and the fitted values are:

I. residuals from the fitted model.
II. in-sample forecast errors.
III. out-of sample forecast errors.

Correct Answer: I and II

Unlike the out-of-sample errors, they are not true ex ante forecasting errors, because the model is typically determined by estimating parameters from all the data in the sample period.