Skip to content

FRM Part I · FRM Exam Part I · Regression Diagnostics

A risk analyst is concerned about severe multicollinearity between two regressors, X1 and X2, in a model used to forecast portfolio losses. Which statement about the consequences and remedies is correct?

With multicollinearity, OLS estimates stay unbiased but their standard errors are inflated, making individual coefficients imprecise. Dropping or combining redundant regressors, or gathering more data, can reduce the problem. It does not cause bias, and weighted least squares is a remedy for heteroskedasticity instead.

  1. AOLS estimators become biased, so the only remedy is to switch to weighted least squares
  2. BThe coefficient estimates remain unbiased but have large standard errors; dropping one of the redundant variables may reduce the problemCorrect
  3. CThe model's R-squared must be low, and adding more correlated variables will improve precision
  4. DPredictions of the dependent variable within the sample range are necessarily unreliable, and the t-tests become too narrow

Explanation

Under imperfect multicollinearity OLS remains unbiased and consistent, but coefficient variances are inflated. Dropping or combining redundant variables, or collecting more data, can help. Weighted least squares addresses heteroskedasticity, not collinearity. R-squared is not necessarily low, and standard errors are too wide, not too narrow.

Did you get it right without looking?

One question tells you little. A timed set on Regression Diagnostics shows your real accuracy, how long you take and where you lose marks.

Practise Regression DiagnosticsFree diagnostic mock

More Regression Diagnostics questions