** Variance Ratio Tests URGENT **

[jugheadj]

Hi all,

I need to perform a test for random walk hypothesis. I know how to carry out the test on e-views however, I'm not sure about some things and I think I keep getting wrong results.

Firstly, I am using stock market returns (not levels). I read somewhere in the forum that we have to use the actual stock market value in the levels instead? (so basically when i plot my returns they are stationary and show time varying volatility, but when i plot the values in the levels it's an increasing trend)

Secondly, when running the test, it gives three options under 'Data Specification' (Random Walk, Exponential Random Walk, and Random walk innovations) - what do these mean and when should I use each of them?

I was trying to replicate Campbell/Lo/Mackinley's results and when I used the CRSP returns (not levels) and chose random walk innovations - I get almost the same results as them? Is this the right way to do it?

I would really appreciate any advice and guidance as I am a beginner with this programme. Thank you :)

[jugheadj]

Anyone? Please?

I think I figured it out but can someone please confirm?
Under "Data Specification":
1) we use Random Walk when our data is in the levels so that Eviews differences the data to perform the VR Test
2) we use Exponential Random Walk when our data is in logs so that Eviews log differences the data to perform the VR Test
3) we use Random Walk Innovations when our data is in returns (i.e. already in the form of P(t)-P(t-1)=m+e where m is drift/mean and e is the innovations)

I tried to replicate Lo/MacKinley's results and when I obtain returns data and use the 3rd option I get pretty similar results.

Can someone please confirm the above? I'd really appreciate it

Thank you!