DCC GARCH model- matrix error in R

Question

I have a fitted DCC GARCH model to daily equity returns of over 70 financial institutions during 2010-2017.I am trying to construct cross-correlations among institutions. However, the code below gives me an error. The problem is I don't know the source of this error.

Below is my code:

rm(list = ls())
options(error=recover)
library(rmgarch)
library(xts)
library(foreign)

dat=read.csv("equity_returns2_2010-17_updated.csv", stringsAsFactors = FALSE)
tickers = unique(dat[,2])
R = do.call(cbind, lapply(1:length(tickers), function(i){
  tmp = dat[which(dat[,2]==tickers[i]),c(1,3)]
  tmp = xts(as.numeric(tmp[,2]), as.Date(strptime(tmp[,1],"%Y%m%d")))
  colnames(tmp)=tickers[i]
  return(tmp)
}))

head(R)
exc=sapply(1:ncol(R), function(i) length(which(is.na(R[,i]))))
exc=which(exc>1)
if(length(exc)>0) R = R[,-exc]
head(R)
which(is.na(R), arr.ind = TRUE)
R = na.omit(R)

m = ncol(R)
spec=ugarchspec(mean.model=list(armaOrder=c(1,0)),variance.model=list(garchOrder=c(1,1),model="gjrGARCH", variance.targeting=TRUE))
spec = multispec(replicate(m, spec))

mspec=dccspec(uspec = spec)
fit = dccfit(mspec, dat=R, solver=c("solnp","solnp"), solver.control=list(trace=1))

it works very well until "fit". Once I run fit it gives the following error:

.
.
.
.
Iter: 1 fn: 2654.2613    Pars:  36.63415  0.97195  0.10379  0.93518 -0.08157
Iter: 2 fn: 2654.2613    Pars:  36.63417  0.97195  0.10379  0.93518 -0.08157
solnp--> Completed in 2 iterations

Iter: 1 fn: 2654.2613    Pars:  36.63415  0.97195  0.10379  0.93518 -0.08157
Iter: 2 fn: 2654.2613    Pars:  36.63417  0.97195  0.10379  0.93518 -0.08157
solnp--> Completed in 2 iterations

error: matrix multiplication: inverse of singular matrix; suggest to use solve() instead

Any idea on how to fix the problem? Thanks in advance.

Answer 1

I dont see any problem with your code. I think you should go beyond the code and look into your data to see whats the problem. have you run this code with any other data set?