*
* Example 6.3.1 from pp 195-196
*
open data ex611.dat
data(format=free,org=columns) 1 200 x
*
@bjident x
set dx = x-x{1}
linreg dx
# constant x{1} dx{1 2}
*
* This will have a slightly different value, because the calculation is being
* done to a higher precision than the one shown in the text, which uses only four
* digits
*
disp "Dickey-Fuller tau statistic" %tstats(2)
*
* The DFUNIT procedure can be employed to do the test as well. The option LAGS
* indicates the number of additional lags of the differenced dependent variable
* to include. Note that the critical values are slightly different from those
* shown in the text. The text is providing asymptotical critical values, while
* the procedure show critical values corrected for sample size.
*
@dfunit(lags=2) x
*
* To do the test without an intercept, use the option DET=NONE. (DET is short for
* DETERMINISTIC). You can also use DET=TREND for the variation of the test which
* includes 1 and time as regressors.
*
@dfunit(lags=2,det=none) x