[Gretl-users] Re: How to apply the HAC option in Gretl?

On Tue, 8 Nov 2022, Sven Schreiber wrote: Granted, there is, but I guess it's better than nothing, provided the user is notified appropriately. In my local tree I patched the source so as to do what I think eviews does; if anyone's curious, the patch is attached. By running the following piece of code <hansl> nulldata 64 set seed 20221109 setobs 1 1 --special-time-series x = uniform() < 0.9 ? normal() : NA y = uniform() < 0.9 ? normal() : NA ols y const x --robust </hansl> I'm getting the following output: <output> Model 1: OLS, using observations 1-64 (T = 53) Missing or incomplete observations dropped: 11 Dependent variable: y HAC standard errors, bandwidth 2 (Bartlett kernel) coefficient std. error t-ratio p-value -------------------------------------------------------- const -0.536282 0.122625 -4.373 6.05e-05 *** x -0.0521759 0.175615 -0.2971 0.7676 Mean dependent var -0.529945 S.D. dependent var 1.023859 Sum squared resid 54.38531 S.E. of regression 1.032656 R-squared 0.002306 Adjusted R-squared -0.017257 F(1, 51) 0.088271 P-value(F) 0.767594 Log-likelihood -75.88750 Akaike criterion 155.7750 Schwarz criterion 159.7156 Hannan-Quinn 157.2904 </output> Sven (or any other): could you cross-check against eviews to see if results match? ------------------------------------------------------- Riccardo (Jack) Lucchetti Dipartimento di Scienze Economiche e Sociali (DiSES) Università Politecnica delle Marche (formerly known as Università di Ancona) r.lucchetti(a)univpm.it http://www2.econ.univpm.it/servizi/hpp/lucchetti -------------------------------------------------------