definition of prediction error

[DanielZ]

Prediction error is defined as process error + parameter error.
Why does this not include model error?

EDIT: I guess you could say this just comes down to a matter of definition, i.e. prediction error is the inherent uncertainty of a prediction given a specific choice of model.

Thanks

 

[zuglubuglu]

What about q&A 4.18i

This question asked to explain the components of the variances of stochastic reserve predictions.

• I argued that firstly there is model error. This can be measured by developing a range of best estimates from different models. Why is this incorrect?
• The model solutions list estimation variance that reflects in perfections derived from the model due to a number of reasons: one of which being the choice of model itself. Isn't this contradictory to the subdivision we make in the notes? Isn't estimation error simply the error derived from choosing incorrect assumptions/parameters?

 

[Darren Michaels]

Have a re-read of Section 3.1 of Chapter 14.

There is sometimes some ambiguity about how these terms are used.

In General:

Prediction variance = Estimation variance + Process variance

Where Estimation variance can be further sub-divided between model error and parameter error.