Econ 140 Final UC Berkeley Fall 2018

Econ 140 Final UC Berkeley Fall 2018 Exam Questions and Answers (Solved Papers) Breusch-Pagan Test [Purpose] - Correct Answers ✅Testing for the presence of heteroskedasticity, and in particular, testing whether the estimated variance of the residuals from a regression is dependent on the values of the independent variables.Weighted Least Squares (WLS) [Purpose] - Correct Answers ✅Transforming the data to adjust for heteroskedasticity, so that the data can use OLS estimators.RESET Test [Purpose] - Correct Answers ✅Detecting omitted variables and incorrect functional form. It is usually sufficient to test for quadratic and cubic terms by generating the squared and cubed versions of the linear prediction, y- hat.Durbin Watson Test [Purpose] - Correct Answers ✅Testing for positive autocorrelation, especially in time-series data Breusch-Pagan Test [Steps] - Correct Answers ✅1. Run the regression of Y on X's

• Generate the predicted residuals, u-hat
• Create u-hat squared
• Run the regression of u-hat squared on X's 1 / 2

Econ 140 Final UC Berkeley Fall 2018 Exam Questions and Answers (Solved Papers)

• Run the joint significance test on the coefficients of the

new regression Breush-Pagan Test [Hypothesis Test Interpretation] - Correct Answers ✅Joint significance test of coefficients - regression of u-hat squared on X's

- H0: gamma1 = gamma2 = .... = gamma(k) = 0

- H1: H0 is false

If we reject H0, then heteroskedasticity exists.If we fail to reject H0, then the data is homoskedastic (ideal case) Weighted Least Squares (WLS) [Steps] - Correct Answers ✅1. Run the regression of Y on X's

• Estimate a model for var(ui|Xi) = h(x) [get the conditional

variance]

• Generate the predicted conditional variance, h(x)-hat
• Create new variables (including the constant), consisting of

the original variables multiplied by 1/√[h(x)-hat]

• Run the regression on weighted Y on weighted X's
• / 2