Author: H. D. Vinod
Date: Sept. 30, 1999

This code is provided gratis without any guarantees or warrantees.
Proprietary modifications of this code are not permitted.
Please make appropriate attribution if you use the code extensively in a research project.

The journal has copyrighted the journal article itself, not my code.

Performance of some GAUSS software routines in terms of numerical accuracy is somewhat disappointing according to H. D. Vinod, "Review of GAUSS for Windows Including Its Numerical Accuracy", Journal of Applied Econometrics, Vol.15, 2000, pp. 211-220. New code with improved numerical accuracy is offered here:

• readme_v.txt
• vinod-progs.zip

The readme file describes the contents. There are many procedures for computing numerically more accurate standard deviation etc. NIST-StRD datasets, their certified correct values, the procedures to compute NADsor number of accurate digits are included. One can learn how to assess the accuracy of GAUSS software by comparing the answers with certified answers. If you have familiarity with GAUSS and some other software X, you can use my methods for testing software X also.

The author provides the following links to the code, which may be occasionally updated:

• http://www.fordham.edu/economics/vinod/readme_v.txt
• http://www.fordham.edu/economics/vinod/vinod-progs.zip

You can also go to
Web page:http://www.fordham.edu/economics/vinod
and then click under Academic: Numerical Accuracy of Gauss Software.

The Journal of Applied Econometrics has also posted the code at a parallel site:
http://qed.econ.queensu.ca/jae/2000-v15.2/vinod/

Hrishikesh D. Vinod
Professor of Economics
E-Mail: Vinod@fordham.edu