Guidance regarding multicollinearity issue

[Umer Aziz]

I hope this email finds you well.

I have transformed all variables into their logarithmic forms. However, I have encountered a multicollinearity issue: one independent variable (IV) and one control variable (CV) exhibit severe multicollinearity, with a VIF value of approximately 47. 

I am uncertain about how to proceed. Removing either variable is not a preferred option, as both are theoretically important to the model. In total, my model includes three independent variables, and the multicollinearity problem is concentrated in one IV and one CV.

Given this situation, I would appreciate your guidance on the following:

1. Are there any alternative methods to address this level of multicollinearity without excluding the variables?

2. Is it acceptable to proceed with the analysis and simply acknowledge the multicollinearity issue in the paper?

3. Since I intend to use the GMM estimator to address endogeneity concerns and conduct Granger causality tests to examine bidirectional relationships, is it appropriate to continue with these analyses despite the high VIF values?

4. Is addressing multicollinearity a necessary prerequisite before proceeding with GMM and Granger causality analysis?

I would be grateful for your advice on the most appropriate course of action.

Thank you for your time and guidance.

Kind Regards 

Research Scholar

[Miklesh Yadav]

Dear Umer,

The following are my inputs:

1) If you think both IDV and control variables are important, and the underlying theory also supports the same, then you may perform principal component analysis of these two variables or residualization to create one variable, and you may use it as IDV. 

2) Acknowledgement of the multicollinearity issue does not serve your purpose; rather, dealing with that issue will serve the purpose. 

3) Please follow point no (1) and go ahead with your analysis. As regards your research objectives (bidirectional causality), you may proceed with Granger causality only; there is no need to proceed with the dynamic panel. However, if you think there is bidirectional causality, and endogeneity occurs due to simultaneity bias, you may proceed with the simultaneous equations model and deal with it with 3SLS (For this, there must be literature support).

(4)  Addressing multicollinearity is not a necessary prerequisite before proceeding with GMM and Granger causality analysis. 

Regards

Dr. Miklesh

--
The members of this group are expected to follow the following Protocols:
1. Please search previous posts in the group before posting the question.
2. Don't write the query in someone's post. Always use the option of New topic for the new question. You can do this by writing to dataanaly...@googlegroups.com
3. It’s better to give a proper subject to your post/query. It'll help others while searching.
4. Never write Open-ended queries. This group intends to help research scholars, NOT TO WORK FOR THEM.
5. Never write words like URGENT in your posts. People will help when they are free.
6. Never upload any information about National Seminars/Conferences. Send such information
in personal emails and feel free to share any RESEARCH-related information.
7. No Happy New Year, Happy Diwali, Happy Holi, Happy Birthday, Happy Anniversary, etc. allowed in this group.
8. Asking or sharing Research Papers is NOT ALLOWED.
9. You can share your questionnaire only once.
---
You received this message because you are subscribed to the Google Groups "DataAnalysis" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dataanalysistrai...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/dataanalysistraining/CAE%3DuFYJD16mK6p9hE8cW%2BARU6mz3hx_Y8KaqTQPheiMZ_w8edw%40mail.gmail.com.