Spurious regression in Stata 1.- Generate two independent unit roots with drift To do this: 1.- set obs 500 2.- generate t=_n 3.- tsset t 4.- gen e1=rnormal() 5.- gen e2=rnormal() 6.- gen y1=e 7.- gen y2=e 8.- replace y1=0.05+L.y1+e1 if t>1 9.- replace y2=0.02+L.y2+e2 if t>1 Plot the time series: graph twoway tsline y1 y2, lpattern(solid dash) regress y1 on y2 to see if they are dependent regress y2 y1, r predict err, resid corrgram err, lags(10) dfuller err, lags(4) regress first difference of y1 on first difference of y2 graph twoway tsline D.y1 D.y2, lpattern(solid dash) regress D.y2 D.y1, r predict err2, resid corrgram err2, lags(10) dfuller err2, lags(4) Note.- Repeat the exercise for two unit roots without drift