[Gretl-users] Re: Newey west standard errors
Hi Ramki,
The textbook standard errors and the ones reported by gretl are both
Newey-West, they just differ in this way: the textbook values have a
degrees-of-freedom adjustment applied while gretl uses the maximum
likelihood estimator of the covariance matrix. You can replicate the
textbook values thus:
ols l_C_expenditure const l_Yd_income l_W_wealth Interest --robust
matrix V = $vcv * $T/($T - $ncoeff)
se = sqrt(diag(V))
print se
Which version is "right"? Well, the rationale for the Newey-West
variance estimator is asymptotic, and opinions vary on whether df
adjustment should be made in such cases.
Allin Cottrell
On Sun, Mar 15, 2026 at 12:50 AM Ramki S <ramakrishnasalagrama(a)gmail.com> wrote:
>
> Dataset:
>
https://raw.githubusercontent.com/srk7774/data/refs/heads/master/table10....
>
>
> Hi,
>
> Is there anyway we can get newey-west standard errors in gretl? HAC robust errors are
differing slightly with textbook result.
>
> Text book result:
> Dependent Variable: LCONSUMPTION Method: Least Squares
> Sample: 1947-2000 Included observations: 54
> Newey-West HAC Standard Errors & Covariance (lag truncation = 3)
>
> Variable Coefficient Std. Error t-Statistic Prob.
> CONSTAN -0.467714 0.043937 -10.64516 0.0000
> LINCOME 0.804871 0.017117 47.02132 0.0000
> LWEALTH 0.201272 0.015447 13.02988 0.0000
> INTEREST -0.002689 0.000880 -3.056306 0.0036
>
> Statistic Value Statistic Value
> R-squared 0.999560 Mean dependent var. 7.826093
> Adjusted R-squared 0.999533 S.D. dependent var. 0.552368
> S.E. of regression 0.011934 F-statistic 37832.71
> Sum squared resid. 0.007121 Prob. (F-statistic) 0.000000
> Durbin-Watson stat. 1.289237
>
> Gretl result:
> Model 10: OLS, using observations 1947-2000 (T = 54)
> Dependent variable: l_C_expenditure
> HAC standard errors, bandwidth 2, Bartlett kernel
>
> coefficient std. error z p-value
> -------------------------------------------------------------
> const −0.467711 0.0458627 −10.20 2.02e-024 ***
> Interest −0.00268906 0.000879808 −3.056 0.0022 ***
> l_Yd_income 0.804873 0.0171749 46.86 0.0000 ***
> l_W_wealth 0.201270 0.0160382 12.55 4.00e-036 ***
>
> Mean dependent var 7.826093 S.D. dependent var 0.552368
> Sum squared resid 0.007121 S.E. of regression 0.011934
> R-squared 0.999560 Adjusted R-squared 0.999533
> F(3, 50) 23576.95 P-value(F) 9.64e-79
> Log-likelihood 164.5880 Akaike criterion −321.1760
> Schwarz criterion −313.2201 Hannan-Quinn −318.1077
> rho 0.324678 Durbin-Watson 1.289219
>
> Log-likelihood for C_expenditure = −258.021
> _______________________________________________
> Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it
> To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it
> Website: https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/