Jahanzeb Malik

Thesis Plan: Using bootstrap methods to discover what is the true model, among a class of similar or related models. Comparing this methodology with existing strategies for model selection. Trying to find out the EFFECTS of model selection on the statistics for the final model selected. Trying to fix the problem that T-statistics are significantly biased upwards when model selection is done prior to estimation. The t-statistics are reported as if it is the FIRST model estimated, without any model selection having been done earlier. It is well known that this creates a bias, but nobody knows how to correct for this bias.

For more details, see my [homepage] -- create a homepage, and put a link here.

Please give your status report here.

AZ: Tasks for Jahanzeb

1. Create your website.

2. Put your SYNOPSIS which was accepted on the main page or some prominent page. Put the bibliography for the synopsis on this page as well. Try to create links to each article so that anyone who wants to can VIEW or DOWNLOAD the article. If the article is attached in PDF format to the website, Google automatically creates a VIEW option, but it does not do this for .DOC files. If the files are large, it may be useful to store them not on the site (which has limited capacity of 100 MB) but on many other free storage places on the web (like www.box.com) and create a link on your website for thedownload.

3. For each of the important articles on the synopsis, create a subpage and indicate the main ideas in this article which related to your research. This will be your annotated bibliography.\

4. See SUBGOALS below

SUBGOALS:

1: Take the results of any recent study and take the two or three most successful lag selection criteria proposed in both contexts. Compare this with your new bootstrap methodology for model selection. Use AR(p) p=1,2,..,5 for example as the class of models to select from, or whatever is used in any recent survey.

2. WRITE a SURVEY of IMPACT of model selection on statistics for final model. Discuss whatever literature exists on the subject -- which is a very small amount of literature if I am not mistake. FOR EACH paper in this literature make sure to put it in your annotated bibliography on your website as discussed above.