Econometrics Review

Flashcards about Econometrics concepts such as Multicollinearity, Heteroscedasticity, Autocorrelation and Simultaneous equations, based on lecture notes.

Multicollinearity

Correlation between explanatory variables in a regression model.

What is Multicollinearity?

When explanatory variables in a regression model are highly correlated.

What happens if there is perfect multicollinearity?

Coefficients of regression of x are indetermined and their standard error is infinite.

What happens if there is almost perfect multicollinearity?

Coefficients of regression of x are determined, but have extremely high standard errors.

How to detect multicollinearity?

R2 > 0.8 and at least one coefficient not significant; R very high.

What is a strategy to deal with multicollinearity?

Eliminate collinear variables or transform the data.

What is Homoscedasticity?

Requires that the variance of the errors be constant across all levels of the independent variables

What are Heteroscedastic errors?

Errors that do not have constant variance

What are causes of heteroscedasticity?

Data itself, outliers, omitted variables

What are the consequences of heteroscedasticity?

Loss of efficiency of OLS estimators, SE are very large, hypothesis tests are invalid, non-optimal decision making

How to detect heteroscedasticity?

Graphical method or statistical approach (Goldfeld-Quandt, Park, Breusch-Pagan, White tests)

What does MCO stand for?

Minimum Ordinary Squares

How to treat heteroscedasticity?

GLS can be used to obtain estimators of the parameters, rescale or transform data, redefine dependent variable

Autocorrelation

A common problem in time series where errors are correlated.

Why does autocorrelation occur?

Observations made at different times are often influenced by prior observations; omit relevant variables; incorrect model specifications; inertia

How to detect autocorrelation?

Graphical method (residuals chart), Durbin-Watson test, Breusch-Godfrey test

What are MES?

Simultaneous Equations Models

What do Simultaneous Equations Models do?

Variables are mutually dependent. Capture interdependence.

How to estimate Simultaneous Equations Models?

Techniques which address endogeneity and interdependence of variables: Two-Stage Least Squares

Logit Model

Estimates the probability of a particular event occurring.