Newey West specifying lag problem

[tomato_wang]

Hello,

I am doing a simple OLS regression. and I found my T-statistics is a little bit big. Then I want to use Newey-West test to adjust the results. I want to use the lag is equal to 12. But When I chose the Newey-West in the estimation options, I could not find how to change the lag. the default lag in the test is 6. so can anybody help me.

[EViews Glenn]

What version of EViews are you using?

[tomato_wang]

I am using Eviews 5 and I also can use Eviews 6 in my univesity libarary.

[tomato_wang]

What version of EViews are you using?

I am using Eviews 5 and I also can use Eviews 6 in my univesity libarary.

[EViews Glenn]

EViews 5 and 6 do not offer the ability in a single equation to specify the kernel bandwidth. You can put your equation into a system object and use the options there to specify the bandwidth.

EViews 7 offers an expanded set of single equation options which allow you to fully specify the kernel bandwidth and/or prewhitening lags.

[porsche_88888]

I also have this problem to do the simple OLS regression specifying lag truncation. In eviews 5.0, when I choose Newey West option, there is no option to set the bandwidth , the lag truncation just calculated automatically instead of be specified by myself.

[tomato_wang]

EViews 5 and 6 do not offer the ability in a single equation to specify the kernel bandwidth. You can put your equation into a system object and use the options there to specify the bandwidth.

EViews 7 offers an expanded set of single equation options which allow you to fully specify the kernel bandwidth and/or prewhitening lags.

Thank you so much. However, I am afraid I could not understand what you mean the system object. Can you tell me how to manipulate the EViews 5 or 6 to specigy the bandwidth with your method?

[EViews Glenn]

It's even a bit more complicated than that as I'll explain in a minute.

Create a new system object by going to the command window and typing system. That will create a new system object. You'll enter your equation there using explicit coefficients.

The complication is that you'll have to estimate this using GMM so you'll need to provide instruments as well. You should enter all of your right-hand side variables as instruments.

I'm not going to go through the full description of this procedure since our manual has a lengthy discussion of systems and system GMM that I would only be repeating. I recommend that you take a look at the system chapter in the manual.

Lastly, I will note that this is easy to do in EViews 7 as we have expanded our tools for single equation covariance estimation considerably. The reason for this is to make problems like this much easier for users to solve...

[tomato_wang]

OK, I'll try the Eviews 7!! thank you so much!