Robust estimation for Seemingly Unrelated Regression Models in presence of cell-wise and case-wise outliers performed using a three-stage procedure. In the first step estimation of the coefficients in each single-equation model is obtained using a Robust Regression procedure, robust estimation of the residual covariance is obtained by a Two-Step Generalized S-estimator, a weighted least square is performed on the whole system to get final estimates of the regression coefficients.
surerob(formula, data, control=lmrob.control(), ...)
# S3 method for surerob
print(x, digits=max(3, getOption("digits")-1), ...)surerob returns a list of the class surerob and
contains all results that belong to the whole system.
This list contains one special object: "eq". It is a list and contains
one object for each estimated equation. These objects are of the class
lmrob and contain the results that belong only to the
regarding equation.
The objects of the class surerob have the following components:
a list that contains the results that belong to the individual equations.
the matched call.
estimation method.
total number of linear independent coefficients.
vector of all estimated coefficients.
matrix of fitted values.
matrix of residuals
imputed residuals from TSGS.
residual covariance matrix used for estimation.
estimated residual covariance matrix.
matrix of robust weights.
object from function TSGS.
list of control parameters used for the estimation.
degrees of freedom of the whole system.
response observations used in the second step.
design matrix used in the second step.
a list of objects of class formula for
multiple-equation models; for single-equation models use function
lmrob.
a list of objects of class data.frame. Each data.frame
contains the data for the corresponding model and all the
data.frames must have the same number of observations.
list of control parameters. The default is constructed
by the function lmrob.control, and it is
passed to function lmrob.
arguments passed to the function
TSGS.
an object of class surerob.
number of digits to print.
Claudio Agostinelli and Giovanni Saraceno
The estimation of systems of equations with unequal numbers of observations is not implemented.
Giovanni Saraceno, Fatemah Alqallaf and Claudio Agostinelli (2021?) A Robust Seemingly Unrelated Regressions For Row-Wise And Cell-Wise Contamination, submitted
lmrob, lm and systemfit
library(systemfit)
data("Kmenta")
eqDemand <- consump~price+income
eqSupply <- consump~price+farmPrice+trend
system <- list(demand=eqDemand, supply=eqSupply)
## Robust estimation
fitrob <- surerob(system, data=Kmenta)
print(fitrob)
Run the code above in your browser using DataLab