101 Concepts for the Level I Exam

Essential Concept 5: Problems in Regression Analysis

Problem  Effect  Solution
Heteroskedasticity: variance of error term is not constant.
Test using BP test
BP = nR^{2}$
F-test is unreliable. Standard error underestimated.
t-stat overstated.
Robust standard errors
Generalized least squares
Serial correlation: error terms are correlated. Test using DW stat.
DW \approx 2(1 – r)
F-stat too high. Standard error underestimated.
t-stat overstated.
Hansen method
Modify the regression equation
Multicollinearity: two or more independent variables are highly correlated.  Inflated SE’s; t-stats of coefficients artificially small
High R^{2}$
Omit one or more of the independent variables