Modeling Risk - Study Session 15

[doobsmeister]

“Modeling risk - the uncertainty in the MBS value that results from the use of assumptions in the complicated Monte Carlo model framework. The MBS value derived from a monte carlo model is very sensitive to the interest rate volatility assumption and the prepayment assumption, for example”
Question: aren’t all models sensitive to this risk of an assumption error? Why is there more error in a monte carlo analysis.
Also on the table in page 298 of shweser - Study session 15, it says that modeling risk is part of the spread. How can this be? Assumptions are inputs and are assumed “in a perfect world.” I don’t understand how this can be part of the spread. Any help would be great here.

 

[S2000magician]

I’ve never heard of modeling risk being limited to Monte Carlo models. You’re correct: all models (of uncertain future events) have modeling risk.
The spread is the buyer’s required return for assuming all risks above those of those of the baseline asset (e.g., a Treasury). Modeling risk is one of those: the risk that the value determined by your model is inaccurate.

 

[doobsmeister]

Thanks Magician, but don’t you think that it is odd that a spread for modeling risk should be added when that risk really depends on the skill of the person building the model? If i created a perfect analyst (in an imaginary world), then this risk should be eliminated.

 

[S2000magician]

It’s a bit odd, and you’re correct: if you could model it perfectly, that risk would be eliminated. As would other risks (e.g., prepayment risk).