Time series - linear vs AR model

CFA Level II
#1

Is my understanding correct that with a linear trend model you can use the durban watson test while with an AR model you have to use a t-test?

#2

i think so

#3

you can use the DW stat in time series as long as the independent variable isn’t a time lagged instance of the dependent variable, this is a very important distinction to make, one that I think will be tested.

#4

Black Swan Wrote: ------------------------------------------------------- > you can use the DW stat in time series as long as > the independent variable isn’t a time lagged > instance of the dependent variable, this is a very > important distinction to make, one that I think > will be tested. you can use the DW stat in regression…

#5

the cfai text explicitely points out that dw can’t be used when the ind VAR is a time lagged instance of the dependent, is it wrong (has happened before) or is my understanding incorrect (happens quite frequently)?

#6

You are right according to the text Swan. DW can’t be used on a AR model. Testing for autocorrelation must be done other ways.