CFA Level II ExamQuantitative MethodsMedium
A quantitative analyst is building a multiple regression model to forecast quarterly GDP growth. The analyst finds that two of the independent variables, interest rates and inflation, exhibit a very high positive correlation (r = 0.92). What is the most likely consequence of this high correlation in the regression model?
- ABias in the intercept term, but not the slope coefficients.
- BReduced overall R-squared of the model.
- CIncreased precision of coefficient estimates.
- DDifficulty in interpreting individual coefficient estimates.
Show answer & explanationAnswer & explanation
Correct answer: D. Difficulty in interpreting individual coefficient estimates.
High correlation between independent variables, known as multicollinearity, makes it difficult to isolate the individual effect of each correlated variable on the dependent variable, leading to unstable and hard-to-interpret coefficient estimates.
Why the other options are wrong
- A. Multicollinearity affects the precision and interpretability of all involved coefficient estimates, including slopes, and can also impact the intercept.
- B. Multicollinearity does not necessarily reduce the overall R-squared; it can even inflate it, though the individual coefficients become unstable.
- C. Multicollinearity generally decreases the precision of coefficient estimates, increasing their standard errors.
Multicollinearity
A phenomenon in multiple regression where two or more independent variables are highly correlated with each other.
- It does not violate OLS assumptions.
- It inflates the standard errors of the coefficient estimates.
- It makes it difficult to interpret the individual impact of correlated variables.
Memory trick: Too much correlation makes coefficients confused.