This preview shows pages 1–3. Sign up to view the full content.
1
VI.
Problems
A. Multicollinearity
B. Heteroskedasticity
C. Autocorrelation
For each “Problem” you should be able to
answer the following questions:
•
What’s the problem?
Define and Explain.
•
What are the consequences of the problem?
•
How do we diagnose the problem?
•
How can we fix it?
VI. Problems
A. Multicollinearity
1. Definition
: The presence of
linear
association
among independent variables.
2. Consequences:
• OLS estimators – remain unbiased.
• Standard errors inflated
⇒
t
calc
deflated.
• Can’t trust your hypothesis tests.
•
P(Type II Error) is high
.
b. Correlation Coefficients
–how strong are
the pair
‐
wise correlations between the Xs?
3. Diagnosis (Multicollinearity)
How can we tell if we have this problem?
a. Classic Signs:
c.
Auxilliary Regressions
–regress one X on all
the others. Is the R2 greater than 0.90?
d. Variance Inflation Factors
(VIF). Same
information as R
2
from the Auxilliary Regs.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document 2
4. Solutions – fixing the problem
•
Sample data problem:
•
Eliminate the offensive variable:
•
Linear Association – try a non
‐
linear model
Linear Association
try a non linear model.
This is the end of the preview. Sign up
to
access the rest of the document.
This note was uploaded on 12/08/2011 for the course ECON 312 taught by Professor Daniellass during the Winter '10 term at UMass (Amherst).
 Winter '10
 DanielLass

Click to edit the document details