GMMStuff/R/iterative.GMM.R
2014-03-03 22:18:26 -05:00

42 lines
1.2 KiB
R

# Iterative GMM method
#
# Author: Francois Pelletier
#
# LGPL-3.0
###############################################################################
#' Iterative GMM method
#'
#' @param start Starting values for the parameters and lagrangian
#' @param conditions.vector Vector of moment conditions
#' @param sample Individual data sample
#' @param ... Functions of the vector of moment conditions
#' @param W Weighting matrix
#' @param R Linear constraint matrix of coefficients
#' @param r Linear constraint constants
#' @param max.iter Maximum number of iterations
#' @param epsilon Minimum precision level
#' @return A list containing the optimized vector of parameter and corresponding covariance matrix
#'
#' @author François Pelletier
iterative.GMM <- function(start,conditions.vector,sample,...,
W,R,r,max.iter=50,epsilon=1E-6)
{
theta1 <- optim.GMM(start,conditions.vector,sample,...,W,R,r)
i <- 1
repeat
{
theta2 <- optim.GMM(theta1,conditions.vector,sample,...,W,R,r)
S <- covariance.GMM(conditions.vector,param,sample,...)
if(sqrt(sum((theta1-theta2)^2))<epsilon)
return(list(par=theta2,cov=S))
else if (i>max.iter)
stop("Iterative GMM does not converge")
else
{
theta1 <- theta2
i <- i+1
}
}
}