A PhD student from Melbourne decided to investigate the extension of opening hours for pubs in Australia in 1966. Some 'concerned' groups thought that longer opening hours would lead to greater consumption of alcohol. He collected quarterly time series data for 44 quarters, 1958Q3-196902. On quantity consumed (Q: defined as retail sales of beer and wine per head of population divided by an index of alcohol prices). He estimated the following relationship: Q+ = 28.60 + 1.325D, - 0.825$1 - 2.775$2 - 2.93653 ratio (54.6) (8.0) (3.8) (13.1) (13.8) R2 = 0.8794 DW = 1.247 N = 44 where the variables are: D -dummy variable that takes the value 1 for years 1966 onwards and 0 otherwise 81= 1 if it is Quarter 1 and 0 otherwise $2 = 1 if it is Quarter 2 and 0 otherwise $3 - 1 if it is Quarter 3 and 0 otherwise DW = Durbin-Watson statistics a) Formally construct hypothesis testing to test the significance of slope coefficients for D and S1 at 5% level of significance. b) Estimate the retail sales of beer and wine in 1967 Quarter 1. c) Examine the hypothesis of is there serially correlated residuals? d) Interpret the implications of autocorrelation for the estimated model above. e) List two alternative ways to overcome autocorrelation