1
$\begingroup$

I need to run Newey West t statistics in SAS 9.2. I already run regression, White's test, Breusch Godfrey test and Jarque-Bera normality test. Regression is simple. Number of observations = 522 Depended variable name: rtest 1 Independent variable name: rtest 2 Data are time series data.

I found somewhere that i should start with:

PROC model;
PARMS B0 B1;
rtest1=B0+B1*rtest2;

I do not understand this part:

FIT rtest1/GMM Kernel=(BART,1,0.605);
RUN;

How to get numbers in parenthesis. What is meaning of those numbers? What does command FIT do to data? Do I need to add more commands?

$\endgroup$
4
  • 1
    $\begingroup$ It's in the manual: look below Figure 18.21. $\endgroup$
    – whuber
    Commented Mar 29, 2011 at 16:13
  • $\begingroup$ @whuber (+1) That is the response, isn't it? $\endgroup$
    – chl
    Commented Mar 29, 2011 at 20:57
  • $\begingroup$ @chl A good response would explain what the Bartlett kernel is and perhaps address the final question about adding more commands. I put this in as a comment really hoping it would result in an improved, more specific question. $\endgroup$
    – whuber
    Commented Mar 29, 2011 at 21:01
  • $\begingroup$ @whuber That's a fair point, indeed. $\endgroup$
    – chl
    Commented Mar 29, 2011 at 21:10

2 Answers 2

3
$\begingroup$

By testing for normality and running more than one test you are not preserving the operating characteristics of the "test". In other words, type I error and confidence interval coverage will be disturbed. A better approach would be to just use a parametric or semi-parametric method and be done with it.

$\endgroup$
0
$\begingroup$

To perform Newey-West standard error correction, PROC MODEL is run again specifying the GMM estimation method in the FIT statement. KERNEL=(BART, 5, 0) is also specified which requests the Bartlett kernel with a lag length of 4. The VARDEF=n option is specified to be consistent with the original Newey-West formula.

 proc model data=two;
        ...
         fit r_invest / gmm kernel=(bart,5,0) vardef=n;
         run;

here is the reference : http://support.sas.com/kb/40/098.html

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.