Heteroskedastic robust t-statistic and F-statistic

[stisen]

Hi everyone. I hope there is someone out there that can answer my question

Im currently writing my thesis about the price-setting on the real estate market, where im making a hedonic model to estimate the price of individual houses, with time on market, physical attributes, locational attributes and trend as independable variables.
I have come to the conclusion that heteroskedasticity is present in my model, why the t-statistic and F-statistic is biased. As far as i know Ramseys RESET test, Wald test and Davidson-MacKinnon test cannot be used with the presence of heteroskedasticity because they rely on the t-statistic and F-statistic. My question is : how can these test be made robust to heteroskedasticity and can it be done in eviews?

My own thought for the Wald test and Davidson-MacKinnon test is first to make the model robust for heteroskedasticity with Whites heteroskedasticy-constant standard error & covariance method and afterwards make the tests. Making the tests without or with Whites heteroskedasticy-constant standard error & covariance, gives different test result, but this doesn't necessarily imply that this is the right way of doing it.
Making the Ramseys RESET test without or with heteroskedasticy-constant standard error & covariance makes no difference in the test results, why i don't think this is the way to go.

Anyway please give me your feedbacks and thoughts on this issue.

Thanks in advance
Martin

[stisen]

There must be someone out there knowing the solution Please share...

[johnrt1987]

Hi,

The procedure you have followed (make estimators robust) I think is correct. When doing the Ramsey test you also have to include the option to make the estimators of the expanded model robust. This will produce a robust F-statistic. I think this is the way.
You can check this book: "Introductory Econometrics, A Modern Approach" by J. Wooldridge. This matter you can find in chapter 9 (third edition).

I hope I have helped somewhat.