Singular Matrix Warning
When you estimate a regression with dummy variables, you have to be careful not to “include the same variable twice.” Try entering g(-1) twice and see what happens:
LS g c g(-1) g(-1)
Here is the message: “Near Singular Matrix”. That is E-views’ way of saying that it is getting a headache trying to carry out the task that you have assigned it. You asked it to determine the separate affects of g(-1) and g(-1) on g. “Huh?” Is E-views’ answer.
Here is another task what will give E-views a headache:
LS g c g(-1) AFTER72 BEFORE72
Again the message: “Near Singular Matrix”. Do you understand why?
Here is another one with quarterly dummies that won’t work:
LS g c g(-1) Q1 Q2 Q3 Q4
Here are some that will work:
LS g g(-1) Q1 Q2 Q3 Q4
LS g c g(-1) Q1 Q2 Q3
Do you understand that both of these are really the same regression, just expressed a little differently??
GAH. They need to improve the help system and UI. Contextual clues as to what a "Near Singular Matrix" is would be nice.
