Hi,
I am currently doing a thesis for my masters.
My research paper is about studying the impact of geopolitical risk in Middle East and North Africa on oil prices.
I wanted to run regressions to test this relationship and study the impact of geopolitical risk in MENA on oil prices and how significant is the relationship and see if the influence is deteriorating over the years.
I was planning on doing a linear regression for this, however, I am not confident on what model is best to use for this and whether OLS will be good enough.
I was wondering what are my options and what would be a viable model to use to study this relationship and test if the relationship is waning.
Many thanks,
Vicken
Depending on your data, (variables) you may experience statistical problems of multiple regression Procedure, so you should avoid most of them before selecting the Regression model which it will be the most appropriate for your study.
Thank you for your response Nazar.
What sort of statistical problems will I encounter and is there a way to avoid them?
I am to regress monthly historical crude oil prices against a single GPR index, while also including multiple control variables.
the best data model for all MENA countries (balanced panel). if you have the right data.
If you use time series Data , (Simple regression ) - you must test for Unit Root .
if you have a Multiple regression - moor than two variables - then you must test for Unit Root before , then you can regress. It could be there are many statistical problems like (Multi correlation, Heterogeneity, Outo correlation).
I will be taking a weighted average of the GPR index of 3 MENA countries (Saudi Arabia, Israel and Turkey) and use this weighted index as a proxy of GPR in MENA. This is because I was only able to find monthly GPR data on those 3 countries but they form a good representation of GPR in MENA.
I aim to have this as my dependent variable and monthly WTI crude oil prices as my independent variable (plus multiple control variables such as US dollar index, industrial production indexes and economic uncertainty). What do you think?
If I test for Unit Root and find that there is Unit Root, what do you think is the best method to tackle the issue? is taking difference of oil prices sufficient? because I tried that but then it complicated regressing oil price returns on GPR as the P-value was too high and so was the F significance.
you can try log oil price returns , or growth rate oil price and test again for UR.
Hi Nazar,
Thank you for being very helpful. I was wondering if you could help me with one more thing.
I want to find a way to find out whether the impact of GPR (independent variables) on oil prices (dependent variable) is increasing or decreasing along the years (my time period is monthly 1985-2019).
What sort of method or test can I deploy to find out whether the relationship between the two variables is deteriorating or not.
The parameter signal gives you the quality of the relationship between the dependent variable and its effect on the dependent variable. The negative means an inverse relationship, and the positive means a positive relationship, taking into account the significance of the real relationship. Greetings to you
- Badges
- Science topic
- Similar topics
- Political Science