Quant Methods - Model specification Errors Table

CFA Level II Quant
#1

Hi All,

I am trying to build a table for quant methods including: Error “name”, how to identify, effect of that violation on the model, which assumption is violtated, etc.

I started with the following, does anyone have a more complete table/explanation and/or could check the following? Thanks!

Heteroskedastic model -> Assumption 4 violated -> Breusch Pagan test -> invalid t stat

Multicolinearity in the model -> Assumption 2 violated -> F is high and t’s are low -> invalid F / t stat

Serial correlation in model -> Assumption 5 violated -> Durbin-Watson test -> Parameters are invalid

Non stationary model -> Unit root test -> If stationary (has unit root), use 1st difference until it becomes stationary

Cointegrated model -> Engel granger test -> if series are cointegrated, model can be used

#2

The F-statistic is unaffected by multicollinearity. Did you see this in a book? Multicollinearity, unless perfect, is not a violation of an assumption and will not impact the F-stat (or R2).