With the sandwich package I want to include a standard robust error in my regression to fix the problem of heteroscedasticity and auto-correlation.
Under the command coeftest, you can select different HC estimators (type = c("HC3", "const", "HC", "HC0", "HC1", "HC2", "HC4", "HC4m", "HC5") for the vcov element.
Does anyone know which HC estimator is most suitable for my regression (see regression below)? or which criteria are used to select the HC estimator correctly?
coeftest(model1.3,vcov. = vcovHAC, type = "HC0")
The dependent variable (giniA) in the regression is a Gini coefficient that can take a value between 0 and 1 (e.g. 0.012, 0.599, 0.987).
The explanatory variables include a dummy variable for the region (West / East) and a dummy variable for the year (2009 / 1999), the gender and age.