Use the data in CONSUMP.XLS and statistical software to run the following regression:

Where RCONPC = real consumer spending in the USA per capita REALYDPC = real disposable income per capita REALR = real interest rate on one-year Treasury Bonds CC = University of Michigan Index of Consumer Confidence (A) Plot the residuals from the regression over time. Does this plot suggest serial correlation is present? Positive or negative?
Explain.
(B) Plot the residuals from the regression against their lagged values. Does this plot suggest serial correlation is present?
Positive or negative? Explain.
(C) Does the regression suffer from serial correlation? Show the five-step procedure of the Durbin-Watson test.
(D) Estimate ρ by regressing the residuals on their lagged values (no constant term). What is the estimate of ρ?
(E) Check for second-order serial correlation. Do you think it is present? Explain.
(F) Estimate ρ and correct for first-order serial correlation using GLS with maximum likelihood estimation. What is this estimate of ρ?
(G) Did GLS remedy the serial correlation? Show the five-step procedure of the Breusch-Godfrey test.
(H) Compare the t-ratios from the original regression and the GLS regression. Are you surprised by which ones turn out to be larger? Explain.
(I) Apply the Newey-West technique to the original regression. Are you surprised at the new standard errors of the estimators?
Explain.

Transcribed Image Text:

RCONPC, B + REALYDPC, + REALR, + BCC, + + e