Hello everyone,
I am running a linear regression and I have to correct my standard errors for serial correlation induced by the overlapping nature of the data. I am using a HAC estimator (quadraticspectral).
As far as I know stata only provides the lag/bandwidth selection according the Newey West's data-based rule. Does anyone of you know how I could determine the bandwidth selection according to Andrews' 1991 data-based rule?
Source: Heteroskedasticity and Autocorrelation Consistent Covariance Matrix. Estimation. Donald W. K. Andrews. Econometrica, Volume 59, Issue 3 (May, 1991)
I am running a linear regression and I have to correct my standard errors for serial correlation induced by the overlapping nature of the data. I am using a HAC estimator (quadraticspectral).
As far as I know stata only provides the lag/bandwidth selection according the Newey West's data-based rule. Does anyone of you know how I could determine the bandwidth selection according to Andrews' 1991 data-based rule?
Source: Heteroskedasticity and Autocorrelation Consistent Covariance Matrix. Estimation. Donald W. K. Andrews. Econometrica, Volume 59, Issue 3 (May, 1991)
Comment