calculate random effects

[Gude]

I am looking to guidance on how the random effects (cross-section, time/balanced, unbalanced) are calculated. I do not mean the model coefficients but the random effects avaiable via View -> Fixed/Random Effects.

I did not find anything in the user's guide nor in text books.

[EViews Glenn]

It's based on the literature from the BLUP (Best Linear Unbiased Predictor) of the component effects. See, for example,
Goldberger (1962) or Baltagi (2001, Econometric Analysis of Panel Data).

[sansa]

Is there any legal version of this books online?

[Gude]

Thank you, Glenn.
Is there any guidance in simpler terms available?

[Gude]

Btw: The "Representations" (View -> Representations) seem a little odd:
Under "Substituted Coefficients", it is suggested to add the random effects to get the fitted values. But one gets the fitted values without adding the random effects. (Note: this is different in the fixed effect setting where one would indeed need to add the fixed effects)

Here is an example for the Grunfeld data in a one-way cross-section RE model:
Substituted Coefficients:
=========================
INV = -57.8460462503 + 0.109783684775*VALUE + 0.308110054652*CAPITAL + [CX=R]


And here for the one-way cross-section FE model:
Substituted Coefficients:
=========================
INV = -59.0109113225 + 0.110124585857*VALUE + 0.310064218491*CAPITAL + [CX=F]

[EViews Glenn]

The representations view is for forecasting. For fitted values, simply subtract the residuals from the observed.