On 11/01/13 18:45, JOSE FRANCISCO PERLES RIBES wrote:
I'm trying to replicate a table similar to that of Angrist and Pischke
(2009) Mostly Harmless Econometrics chapter 8 p,306 where in a Monte
Carlo experiment comparing different bootstrap standard errors
(conventional versus White, ...HAC) for a regression coefficient of
interest, to show that the robust errors tend to be biased if the
problem that are trying to solve (heteroskedasticity or
autocorrelation) is not very serious.
I have seen that Gretl GUI has a bootstrap option when estimates a
model, but it is for the confidence interval and p-value. This option
allows you save the values, which is very interesting to get the mean
and standard deviation of the beta coefficient.
But my question is, there is no option to save the estimated standard
errors directly,and via descriptive statistics to obtain the mean and
standard deviation of these errors?
Thanks in advance and sorry for any inconvenience.
José F. Perles Ribes
Applied Economic Analysis Department
University of Alicante (Spain)
You cannot do it via the gretl GUI. You need to program it into a
script. Look at the manual for this, section 6.4 "Resampling and
DEPARTAMENTO DE ECONOMÍA APLICADA III (ECONOMETRÍA Y ESTADÍSTICA)
UPV/EHU Avda. Lehendakari Aguirre, 83 | 48015 BILBAO
T.: +34 946013732 | F.: +34 946013754