Background: I'm interested in the econometrics behind identifying if a function of the nature y= min{x1,x2} exists in a given DGP . I'm wondering if multicollinearity of variables has something to do with it.
Suppose you have a regression:
y=b0+b1x1+b2x2+u
where
1) x1 and x2 are correlated with each other
2) x1 and x2 are qualitatively different variables.
can one suppose that some function of the nature min{x1,x2} is present in the determination of y?
Dear Jacob Smith,
Multicollinearity refers to the linear correlation between two or more explanatory variables. My humble suggestion is that it does not suggest that two variables are complementary in determining the DV. According to Gujrati (2013), the multicollinearity is not an matter of presence of absence, its severity matters. Generally, a value of VIF (variance inflating factor) less than 5 is recommended and you can find a lot of literature support for that too. Any further discussion on the issue is welcomed.
Best wishes,
Atif
Dear Jacob Smith,
Multicollinearity refers to the linear correlation between two or more explanatory variables. My humble suggestion is that it does not suggest that two variables are complementary in determining the DV. According to Gujrati (2013), the multicollinearity is not an matter of presence of absence, its severity matters. Generally, a value of VIF (variance inflating factor) less than 5 is recommended and you can find a lot of literature support for that too. Any further discussion on the issue is welcomed.
Best wishes,
Atif
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression