Multicollinearity
This type of multicollinearity occurs when explanatory variables are completely linearly related
What is perfect multicollinearity
This type of heteroscedasticity arises only from the error term of a correctly specified model.
What is pure heteroscedasticity?
A DW statistic close to 2 indicates what?
What is no autocorrelation?
Omitting a relevant variable, such as dropping X3 from a cubic cost function, leads to what type of specification error?
What is underfitting?
A lin-log model is written as Y=β0+β1ln(X)+u. What kind of change in Y does β1 measure?
What is absolute change in Y for a % change in X?
Name one practical consequence of multicollinearity, such as high R² but low t-stats or unstable coefficients.
What is large SEs / small t-values, wide CIs, wrong coefficient signs, or unstable estimates?
Confidence intervals and t-tests become unreliable in the presence of heteroscedasticity because these become incorrect.
What are standard errors?
In cross-sectional data, autocorrelation is usually called this.
What is spatial autocorrelation?
Even when omitted and included variables are uncorrelated, which parameter remains biased?
What is the intercept?
A log-lin model is written as ln(Y)=β0+β1T+u. What does β1 represent?
What is the growth rate of Y?
This statistic measures how much the variance of a coefficient is inflated due to collinearity.
What is the Variance Inflation Factor (VIF)
What is the main danger heteroscedasticity poses for hypothesis testing?
What is misleading t-stats and F-tests?
Autocorrelation caused by misspecification (e.g., omitted variable or incorrect functional form) is called what?
What is impure serial correlation?
Fitting a log-linear model when the true relationship is linear is an example of what specification error?
What is incorrect functional form?
The slope coefficient β11 in a log-log model measures what economically meaningful concept?
What is elasticity?
In the presence of near multicollinearity, OLS estimators are still BLUE, but this aspect becomes very large, making coefficients imprecise.
What is large standard errors.
The most widely used remedy leaves the coefficient estimates unchanged but corrects the standard errors. What is it?
What are heteroscedasticity-corrected (HC/White) standard errors?
In the presence of positive autocorrelation, the usual estimator of σ2 is likely to do what?
What is underestimate the true variance?
Ramsey’s RESET test detects what types of model problems?
What are omitted variables, incorrect functional form, or neglected nonlinearities?
Why can’t you compare R² values across models with different dependent variables?
Because R² depends on the scale of the response variable.
This regression variant intentionally introduces bias to reduce coefficient variance under multicollinearity.
What is ridge regression?
When heteroscedasticity appears proportional to Xi, this transformation is recommended.
What is the square-root transformation (Yi/Xi)?
The Durbin–Watson test is invalid if the regression model contains what type of term?
What is a lagged dependent variable (e.g.,Yt-1)?
Mallows Cp is ideal when its value is approximately equal to what?
What is the number of predictors plus one (p+1)?
In a log-log model, the slope coefficient is 0.6. If X increases by 10%, what is the predicted percentage change in Y?
What is a 6% increase in Y?