log the price index before unit root testing or not?

[peter_h]

hey all,

i know that when dealt with price indices we take a natural logs before estimating a model. But I'm wondering if we should log the price indices ALSO before testing for unit roots. I used all three major tests (ADF, PP, KPSS) and some of the results are different before and after taking logs. I am aware that theory suggests that price indices are I(1). Can you give me a hint? thank a lot.

[startz]

hey all,

i know that when dealt with price indices we take a natural logs before estimating a model. But I'm wondering if we should log the price indices ALSO before testing for unit roots. I used all three major tests (ADF, PP, KPSS) and some of the results are different before and after taking logs. I am aware that theory suggests that price indices are I(1). Can you give me a hint? thank a lot.

The best guide is probably to apply the test to whichever variable (level or log) tha you are going to use in your analysis.

[adampc]

Hi,

I basically have a similar question.

I have equity returns which I wish to regress on a quarterly basis. So I have quarterly returns for X (e.g. +2.5%, -3.4%, -0.2%, +5,7%) and for Y.

My question is should I regress the quarterly returns against each other or shall I rebase it in an index form, e.g. starts at 100, then 102.5, 99.8, etc?

Finally, should I log any of this data?

Thanks