STATE CALIFORNIA UNIVERSITY, HAYWARD DEPARTMENT OF STATISTICS ECOMOMICS 6400 Seminar in Econometrics Examples: Serial Correlation Example 1: Read Example 9.1 pages 381-382 and Example 9.2 pages 387-388, open the dataset DATA6-6, and run the Practice Computer Session 9.1 Open dataset data6-6 File > Open data > sample file > Ramanathan double click on Data6-6 U.S. farm population View the data. Data > Display variables > all variables Open practice file ps9-1 File > Open command file > practice file > Ramanathan double click on Ps9-1 autocorrelation Start a gretl console Utilities gretl console First, plot the data by typing the following command at the gretl prompt ? ? graph farmpop time Does there seem to be autocorrelation in the data? Answer: Yes, since there is a trend in the data due to time. Copy (Ctrl-C) and Paste (Ctrl-V) each command line from the practice file ps9-1 into the gretl console and see what each command line produces. population
genr time
(* creates the time variable farmpop *)
ols const time genr yhat=farmpop-uhat genr ut=uhat genr ut1=ut(-1) corr ut ut1 *) graph ut time
(* runs the ols y = farmpop, x = time *) (* saves the fitted values, yhat *) (* saves the estimated residuals *) (* saves the lagged residuals *) (* computes the estimated autocorrelation in the residuals (* plots the residual plot *)
In the example the autocorrelation in the farmpop data is shown and estimated. Perform the DW test to confirm the presence of autocorrelation. Example 2: Read Example 9.3 page 388 and Example 9.6 page 394, open the dataset DATA4-7, run the Practice Computer Session 9.2 ...